[client/ui] Update Wails to v3.0.0-alpha2.105

Bump github.com/wailsapp/wails/v3 to v3.0.0-alpha2.105 and the
@wailsio/runtime frontend dependency to 3.0.0-alpha.94 (latest published
on npm), excluding it from the pnpm minimum release age gate.

.coderabbit.yaml

@@ -0,0 +1,18 @@
+# yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json
+language: en-US
+reviews:
+  profile: chill
+  request_changes_workflow: false
+  high_level_summary: true
+  poem: false
+  review_status: true
+  auto_review:
+    enabled: true
+    drafts: false
+  path_filters:
+    - "!**/*.tsx"
+    - "!**/*.ts"
+    - "!**/*.js"
+    - "!**/*.svg"
+chat:
+  auto_reply: true

.devcontainer/Dockerfile

@@ -6,7 +6,6 @@ RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
         iptables=1.8.9-2 \
         libgl1-mesa-dev=22.3.6-1+deb12u1 \
         xorg-dev=1:7.7+23 \
-        libayatana-appindicator3-dev=0.5.92-1 \
     && apt-get clean \
     && rm -rf /var/lib/apt/lists/* \
     && go install -v golang.org/x/tools/gopls@latest

.github/workflows/check-license-dependencies.yml

@@ -20,7 +20,7 @@ jobs:
 
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
@@ -59,12 +59,12 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Set up Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: true

.github/workflows/frontend-ui.yml

@@ -0,0 +1,98 @@
+name: UI Frontend
+
+on:
+  pull_request:
+    paths:
+      - "client/ui/frontend/**"
+      - "client/ui/i18n/**"
+      - "client/ui/**/*.go"
+      - ".github/workflows/frontend-ui.yml"
+  push:
+    branches:
+      - main
+    paths:
+      - "client/ui/frontend/**"
+      - "client/ui/i18n/**"
+      - "client/ui/**/*.go"
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.actor_id }}
+  cancel-in-progress: true
+
+jobs:
+  lint-and-build:
+    name: Lint & Build
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+    defaults:
+      run:
+        working-directory: client/ui/frontend
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          persist-credentials: false
+
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: "22"
+
+      - name: Set up pnpm
+        uses: pnpm/action-setup@a3252b78c470c02df07e9d59298aecedc3ccdd6d # v3.0.0
+        with:
+          version: 11
+
+      # Bindings are generated by wails3 from the Go service definitions and
+      # are not checked in (see client/ui/frontend/bindings/). Without them,
+      # typecheck/build fail on missing module imports.
+      - name: Set up Go
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        with:
+          go-version-file: "go.mod"
+          cache: false
+
+      # wails3 CLI links against GTK4 / WebKitGTK 6.0 via its internal/operatingsystem
+      # package, so the dev libraries must be present before `go install`.
+      - name: Install Wails Linux system dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            pkg-config \
+            libgtk-4-dev \
+            libwebkitgtk-6.0-dev
+
+      - name: Install wails3 CLI
+        # Version derived from go.mod so the binding generator always matches
+        # the wails runtime the daemon links against.
+        working-directory: ${{ github.workspace }}
+        run: |
+          WAILS_VERSION=$(go list -m -f '{{.Version}}' github.com/wailsapp/wails/v3)
+          go install github.com/wailsapp/wails/v3/cmd/wails3@$WAILS_VERSION
+
+      - name: Get pnpm store directory
+        id: pnpm-store
+        run: echo "path=$(pnpm store path --silent)" >> "$GITHUB_OUTPUT"
+
+      - name: Cache pnpm store
+        uses: actions/cache@v4
+        with:
+          path: ${{ steps.pnpm-store.outputs.path }}
+          key: ${{ runner.os }}-pnpm-${{ hashFiles('client/ui/frontend/pnpm-lock.yaml') }}
+          restore-keys: |
+            ${{ runner.os }}-pnpm-
+
+      - name: Install dependencies
+        run: pnpm install --frozen-lockfile
+
+      - name: Generate Wails bindings
+        run: pnpm run bindings
+
+      - name: Lint, typecheck, format
+        run: pnpm check
+
+      - name: Build
+        run: pnpm build

.github/workflows/git-town.yml

@@ -15,7 +15,7 @@ jobs:
       pull-requests: write
 
     steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - uses: git-town/action@3d8b878379abb1ee393fb49865a28b4a6c2cd3b0 # v1.2.1

.github/workflows/golang-test-darwin.yml

@@ -16,18 +16,18 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
 
       - name: Cache Go modules
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: ~/go/pkg/mod
           key: macos-gotest-${{ hashFiles('**/go.sum') }}
@@ -45,10 +45,18 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Test
-        run: NETBIRD_STORE_ENGINE=${{ matrix.store }} CI=true go test -coverprofile=coverage.txt -tags 'devcert privileged' -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE' -timeout 5m -p 1 $(go list ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined -e /client/testutil/privileged)
+        # Exclude client/ui: its main.go uses //go:embed all:frontend/dist,
+        # which fails to compile until the frontend has been built. The Wails UI
+        # has no Go-side unit tests, and its release pipeline runs `pnpm build`
+        # before goreleaser.
+        # `go list -e` lets the listing succeed even though the embed fails to
+        # resolve; the grep then drops the broken package by path. Without -e,
+        # go list aborts with empty stdout and `go test` falls back to the repo
+        # root, which has no Go files.
+        run: NETBIRD_STORE_ENGINE=${{ matrix.store }} CI=true go test -coverprofile=coverage.txt -tags=devcert -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE' -timeout 5m -p 1 $(go list -e ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined -e /client/ui)
 
       - name: Upload coverage reports to Codecov
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird

.github/workflows/golang-test-freebsd.yml

@@ -16,7 +16,7 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
@@ -28,7 +28,7 @@ jobs:
         id: test
         env:
           GO_VERSION: ${{ steps.goversion.outputs.version }}
-        uses: vmactions/freebsd-vm@b84ab5559b5a1bb4b8ee2737d2506a16e1737636 # v1.4.8
+        uses: vmactions/freebsd-vm@d1e65811565151536c0c894fff74f06351ed26e6 # v1.4.5
         with:
           usesh: true
           copyback: false
@@ -48,14 +48,14 @@ jobs:
             export PATH=$PATH:/usr/local/go/bin:$HOME/go/bin
             time go build -o netbird client/main.go
             # check all component except management, since we do not support management server on freebsd
-            time go test -tags privileged -timeout 1m -failfast ./base62/...
+            time go test -timeout 1m -failfast ./base62/...
             # NOTE: without -p1 `client/internal/dns` will fail because of `listen udp4 :33100: bind: address already in use`
-            time go test -tags privileged -timeout 8m -failfast -v -p 1 ./client/...
-            time go test -tags privileged -timeout 1m -failfast ./dns/...
-            time go test -tags privileged -timeout 1m -failfast ./encryption/...
-            time go test -tags privileged -timeout 1m -failfast ./formatter/...
-            time go test -tags privileged -timeout 1m -failfast ./client/iface/...
-            time go test -tags privileged -timeout 1m -failfast ./route/...
-            time go test -tags privileged -timeout 1m -failfast ./sharedsock/...
-            time go test -tags privileged -timeout 1m -failfast ./util/...
-            time go test -tags privileged -timeout 1m -failfast ./version/...
+            time go test -timeout 8m -failfast -v -p 1 ./client/...
+            time go test -timeout 1m -failfast ./dns/...
+            time go test -timeout 1m -failfast ./encryption/...
+            time go test -timeout 1m -failfast ./formatter/...
+            time go test -timeout 1m -failfast ./client/iface/...
+            time go test -timeout 1m -failfast ./route/...
+            time go test -timeout 1m -failfast ./sharedsock/...
+            time go test -timeout 1m -failfast ./util/...
+            time go test -timeout 1m -failfast ./version/...

.github/workflows/golang-test-linux.yml

@@ -18,7 +18,7 @@ jobs:
       management: ${{ steps.filter.outputs.management }}
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
@@ -30,7 +30,7 @@ jobs:
               - 'management/**'
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -41,7 +41,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         id: cache
         with:
           path: |
@@ -53,7 +53,7 @@ jobs:
 
       - name: Install dependencies
         if: steps.cache.outputs.cache-hit != 'true'
-        run: sudo apt update && sudo apt install -y -q libgtk-3-dev libayatana-appindicator3-dev libgl1-mesa-dev xorg-dev gcc-multilib libpcap-dev
+        run: sudo apt update && sudo apt install -y -q libgtk-4-dev libwebkitgtk-6.0-dev libsoup-3.0-dev libgl1-mesa-dev xorg-dev gcc-multilib libpcap-dev
 
       - name: Install 32-bit libpcap
         if: steps.cache.outputs.cache-hit != 'true'
@@ -119,12 +119,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -135,7 +135,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -145,7 +145,7 @@ jobs:
             ${{ runner.os }}-gotest-cache-
 
       - name: Install dependencies
-        run: sudo apt update && sudo apt install -y -q libgtk-3-dev libayatana-appindicator3-dev libgl1-mesa-dev xorg-dev gcc-multilib libpcap-dev
+        run: sudo apt update && sudo apt install -y -q libgtk-4-dev libwebkitgtk-6.0-dev libsoup-3.0-dev libgl1-mesa-dev xorg-dev gcc-multilib libpcap-dev
 
       - name: Install 32-bit libpcap
         if: matrix.arch == '386'
@@ -158,29 +158,36 @@ jobs:
         run: git --no-pager diff --exit-code
 
       - name: Test
-        run: CGO_ENABLED=1 GOARCH=${{ matrix.arch }} CI=true go test -coverprofile=coverage.txt -tags devcert -timeout 10m -p 1 $(go list ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined)
+        # Exclude client/ui: its main.go uses //go:embed all:frontend/dist,
+        # which fails to compile until the frontend has been built. The Wails UI
+        # has no Go-side unit tests, and its release pipeline runs `pnpm build`
+        # before goreleaser.
+        # `go list -e` lets the listing succeed even though the embed fails to
+        # resolve; the grep then drops the broken package by path. Without -e,
+        # go list aborts with empty stdout and `go test` falls back to the repo
+        # root, which has no Go files.
+        run: CGO_ENABLED=1 GOARCH=${{ matrix.arch }} CI=true go test -coverprofile=coverage.txt -tags devcert -exec 'sudo' -timeout 10m -p 1 $(go list -e ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined -e /client/ui)
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
           flags: unit,client
 
-
   test_client_on_docker:
     name: "Client (Docker) / Unit"
     needs: [build-cache]
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -192,7 +199,7 @@ jobs:
           echo "modcache_dir=$(go env GOMODCACHE)" >> $GITHUB_OUTPUT
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         id: cache-restore
         with:
           path: |
@@ -229,7 +236,7 @@ jobs:
             sh -c ' \
               apk update; apk add --no-cache \
                 ca-certificates iptables ip6tables dbus dbus-dev libpcap-dev build-base; \
-              go test -buildvcs=false -tags "devcert privileged" -v -timeout 10m -p 1 $(go list -buildvcs=false ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined -e /client/ui -e /upload-server -e /client/testutil/privileged)
+              go test -buildvcs=false -tags devcert -v -timeout 10m -p 1 $(go list -e -buildvcs=false ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined -e /client/ui -e /upload-server)
             '
 
   test_relay:
@@ -246,12 +253,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -266,7 +273,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -290,7 +297,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -306,12 +313,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -325,7 +332,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -347,7 +354,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -363,12 +370,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -383,7 +390,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -407,7 +414,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -424,12 +431,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -440,7 +447,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -484,7 +491,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -529,12 +536,12 @@ jobs:
             prom/prometheus
 
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -545,7 +552,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -579,11 +586,10 @@ jobs:
           CGO_ENABLED=1 GOARCH=${{ matrix.arch }} \
           NETBIRD_STORE_ENGINE=${{ matrix.store }} \
           CI=true \
+          GIT_BRANCH=${{ github.ref_name }} \
           go test -tags devcert -run=^$ -bench=. \
           -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE,GIT_BRANCH,GITHUB_RUN_ID' \
           -timeout 20m ./management/... ./shared/management/... $(go list ./management/... ./shared/management/... | grep -v -e /management/server/http)
-        env:
-          GIT_BRANCH: ${{ github.ref_name }}
 
   api_benchmark:
     name: "Management / Benchmark (API)"
@@ -624,12 +630,12 @@ jobs:
             prom/prometheus
 
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -640,7 +646,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -674,13 +680,12 @@ jobs:
           CGO_ENABLED=1 GOARCH=${{ matrix.arch }} \
           NETBIRD_STORE_ENGINE=${{ matrix.store }} \
           CI=true \
+          GIT_BRANCH=${{ github.ref_name }} \
           go test -tags=benchmark \
             -run=^$ \
             -bench=. \
             -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE,GIT_BRANCH,GITHUB_RUN_ID' \
             -timeout 20m ./management/server/http/...
-        env:
-          GIT_BRANCH: ${{ github.ref_name }}
 
   api_integration_test:
     name: "Management / Integration"
@@ -694,12 +699,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -710,7 +715,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -736,7 +741,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
+        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird

.github/workflows/golang-test-windows.yml

@@ -18,12 +18,12 @@ jobs:
     runs-on: windows-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         id: go
         with:
           go-version-file: "go.mod"
@@ -35,7 +35,7 @@ jobs:
           echo "modcache=$(go env GOMODCACHE)" >> $env:GITHUB_ENV
 
       - name: Cache Go modules
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ${{ env.cache }}
@@ -65,10 +65,17 @@ jobs:
       - run: PsExec64 -s -w ${{ github.workspace }} C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe env -w GOCACHE=${{ env.modcache }}
       - run: PsExec64 -s -w ${{ github.workspace }} C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe mod tidy
       - name: Generate test script
+        # Exclude client/ui: its main.go uses //go:embed all:frontend/dist,
+        # which fails to compile until the frontend has been built. The Wails UI
+        # has no Go-side unit tests, and its release pipeline runs `pnpm build`
+        # before goreleaser.
+        # `go list -e` lets the listing succeed even though the embed fails to
+        # resolve; the Where-Object pipeline then drops the broken package by
+        # path. Without -e, go list aborts with empty stdout.
         run: |
-          $packages = go list ./... | Where-Object { $_ -notmatch '/management' } | Where-Object { $_ -notmatch '/relay' } | Where-Object { $_ -notmatch '/signal' } | Where-Object { $_ -notmatch '/proxy' } | Where-Object { $_ -notmatch '/combined' }
+          $packages = go list -e ./... | Where-Object { $_ -notmatch '/management' } | Where-Object { $_ -notmatch '/relay' } | Where-Object { $_ -notmatch '/signal' } | Where-Object { $_ -notmatch '/proxy' } | Where-Object { $_ -notmatch '/combined' } | Where-Object { $_ -notmatch '/client/ui' }
           $goExe = "C:\hostedtoolcache\windows\go\${{ steps.go.outputs.go-version }}\x64\bin\go.exe"
-          $cmd = "$goExe test -tags `"devcert privileged`" -timeout 10m -p 1 $($packages -join ' ') > test-out.txt 2>&1"
+          $cmd = "$goExe test -tags=devcert -timeout 10m -p 1 $($packages -join ' ') > test-out.txt 2>&1"
           Set-Content -Path "${{ github.workspace }}\run-tests.cmd" -Value $cmd
 
       - name: test

.github/workflows/golangci-lint.yml

@@ -15,14 +15,22 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - name: codespell
         uses: codespell-project/actions-codespell@8f01853be192eb0f849a5c7d721450e7a467c579 # v2.2
         with:
           ignore_words_list: erro,clienta,hastable,iif,groupd,testin,groupe,cros,ans,deriver,te,userA,ede,additionals
-          skip: go.mod,go.sum,**/proxy/web/**
+          # Non-English UI translations trip codespell on real foreign words
+          # (de: "Sie", "oder", "ist"). Only en/common.json is the source of
+          # truth that should be spell-checked. List each translated locale
+          # dir below and add new ones as languages are added under
+          # client/ui/i18n/locales/. Single-star globs are matched per path
+          # segment by codespell and behave the same across versions; the
+          # recursive "**" form did not take effect with the codespell shipped
+          # by this action.
+          skip: go.mod,go.sum,*/proxy/web/*,*pnpm-lock.yaml,*package-lock.json,*/locales/de/*,*/locales/es/*,*/locales/fr/*,*/locales/hu/*,*/locales/it/*,*/locales/pt/*,*/locales/ru/*,*/locales/zh-CN/*,*/i18n/TRANSLATING.md
   golangci:
     strategy:
       fail-fast: false
@@ -40,7 +48,7 @@ jobs:
     timeout-minutes: 15
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - name: Check for duplicate constants
@@ -48,13 +56,22 @@ jobs:
         run: |
           ! awk '/const \(/,/)/{print $0}' management/server/activity/codes.go | grep -o '= [0-9]*' | sort | uniq -d | grep .
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
       - name: Install dependencies
         if: matrix.os == 'ubuntu-latest'
-        run: sudo apt update && sudo apt install -y -q libgtk-3-dev libayatana-appindicator3-dev libgl1-mesa-dev xorg-dev libpcap-dev
+        run: sudo apt update && sudo apt install -y -q libgtk-4-dev libwebkitgtk-6.0-dev libsoup-3.0-dev libgl1-mesa-dev xorg-dev libpcap-dev
+      - name: Stub Wails frontend bundle
+        # client/ui/main.go has //go:embed all:frontend/dist. The
+        # directory is produced by `pnpm run build` and is gitignored, so
+        # lint-only runs (no frontend toolchain) need a placeholder file
+        # for the embed pattern to match.
+        shell: bash
+        run: |
+          mkdir -p client/ui/frontend/dist
+          touch client/ui/frontend/dist/.embed-placeholder
       - name: golangci-lint
         uses: golangci/golangci-lint-action@82606bf257cbaff209d206a39f5134f0cfbfd2ee #v9.2.1
         with:

.github/workflows/install-script-test.yml

@@ -22,7 +22,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     steps:
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 

.github/workflows/mobile-build-validation.yml

@@ -16,11 +16,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
       - name: Setup Android SDK
@@ -28,13 +28,13 @@ jobs:
         with:
           cmdline-tools-version: 8512546
       - name: Setup Java
-        uses: actions/setup-java@1bcf9fb12cf4aa7d266a90ae39939e61372fe520
+        uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654
         with:
           java-version: "11"
           distribution: "adopt"
       - name: NDK Cache
         id: ndk-cache
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: /usr/local/lib/android/sdk/ndk
           key: ndk-cache-23.1.7779620
@@ -54,11 +54,11 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
       - name: install gomobile

.github/workflows/release.yml

@@ -9,7 +9,7 @@ on:
   pull_request:
 
 env:
-  SIGN_PIPE_VER: "v0.1.6"
+  SIGN_PIPE_VER: "v0.1.8"
   GORELEASER_VER: "v2.16.0"
   PRODUCT_NAME: "NetBird"
   COPYRIGHT: "NetBird GmbH"
@@ -27,7 +27,7 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
@@ -64,7 +64,7 @@ jobs:
         if: steps.check_diff.outputs.diff_exists == 'true'
         env:
           GO_VERSION: ${{ steps.goversion.outputs.version }}
-        uses: vmactions/freebsd-vm@b84ab5559b5a1bb4b8ee2737d2506a16e1737636 # v1.4.8
+        uses: vmactions/freebsd-vm@d1e65811565151536c0c894fff74f06351ed26e6 # v1.4.5
         with:
           usesh: true
           copyback: false
@@ -135,7 +135,7 @@ jobs:
       ghcr_images: ${{ steps.tag_and_push_images.outputs.images_markdown }}
     steps:
       - name: Checkout
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0 # It is required for GoReleaser to work properly
           persist-credentials: false
@@ -166,12 +166,12 @@ jobs:
           fi
 
       - name: Set up Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
       - name: Cache Go modules
-        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache/restore@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ~/go/pkg/mod
@@ -186,9 +186,9 @@ jobs:
       - name: check git status
         run: git --no-pager diff --exit-code
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@06116385d9baf250c9f4dcb4858b16962ea869c3 #v4.1.0
+        uses: docker/setup-qemu-action@ce360397dd3f832beb865e1373c09c0e9f86d70a #v4.0.0
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@d7f5e7f509e45cec5c76c4d5afdd7de93d0b3df5 #v4.1.0
+        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd #v4.0.0
       - name: Login to Docker hub
         if: github.event_name != 'pull_request'
         uses: docker/login-action@650006c6eb7dba73a995cc03b0b2d7f5ca915bee # v4.2.0
@@ -216,12 +216,12 @@ jobs:
       - name: Install goversioninfo
         run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
       - name: Generate windows syso amd64
-        run: goversioninfo  -icon client/ui/assets/netbird.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_amd64.syso
+        run: goversioninfo  -icon client/ui/build/windows/icon.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_amd64.syso
       - name: Generate windows syso arm64
-        run: goversioninfo -arm -64 -icon client/ui/assets/netbird.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_arm64.syso
+        run: goversioninfo -arm -64 -icon client/ui/build/windows/icon.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_arm64.syso
       - name: Run GoReleaser
         id: goreleaser
-        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
+        uses: goreleaser/goreleaser-action@4c6ab561adb47e50c45ef534e2155934e91c40c1 # v7.2.0
         with:
           version: ${{ env.GORELEASER_VER }}
           args: release --clean ${{ env.flags }}
@@ -347,7 +347,7 @@ jobs:
       release_ui_artifact_url: ${{ steps.upload_release_ui.outputs.artifact-url }}
     steps:
       - name: Checkout
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0 # It is required for GoReleaser to work properly
           persist-credentials: false
@@ -374,12 +374,12 @@ jobs:
           fi
 
       - name: Set up Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
       - name: Cache Go modules
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ~/go/pkg/mod
@@ -394,8 +394,18 @@ jobs:
       - name: check git status
         run: git --no-pager diff --exit-code
 
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+
+      - name: Set up pnpm
+        uses: pnpm/action-setup@a3252b78c470c02df07e9d59298aecedc3ccdd6d # v3.0.0
+        with:
+          version: 11
+
       - name: Install dependencies
-        run: sudo apt update && sudo apt install -y -q libappindicator3-dev gir1.2-appindicator3-0.1 libxxf86vm-dev gcc-mingw-w64-x86-64
+        run: sudo apt update && sudo apt install -y -q libgtk-4-dev libwebkitgtk-6.0-dev libsoup-3.0-dev gcc-mingw-w64-x86-64
 
       - name: Decode GPG signing key
         if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
@@ -414,13 +424,19 @@ jobs:
           echo "/tmp/llvm-mingw-20250709-ucrt-ubuntu-22.04-x86_64/bin" >> $GITHUB_PATH
       - name: Install goversioninfo
         run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
+      - name: Install wails3 CLI
+        # Version derived from go.mod so the binding generator always matches
+        # the wails runtime the binary links against.
+        run: |
+          WAILS_VERSION=$(go list -m -f '{{.Version}}' github.com/wailsapp/wails/v3)
+          go install github.com/wailsapp/wails/v3/cmd/wails3@$WAILS_VERSION
       - name: Generate windows syso amd64
-        run: goversioninfo -64 -icon client/ui/assets/netbird.ico -manifest client/ui/manifest.xml -product-name ${{ env.PRODUCT_NAME }}-"UI" -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/ui/resources_windows_amd64.syso
+        run: goversioninfo -64 -icon client/ui/build/windows/icon.ico -manifest client/ui/build/windows/wails.exe.manifest -product-name ${{ env.PRODUCT_NAME }}-"UI" -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/ui/resources_windows_amd64.syso
       - name: Generate windows syso arm64
-        run: goversioninfo -arm -64 -icon client/ui/assets/netbird.ico -manifest client/ui/manifest.xml -product-name ${{ env.PRODUCT_NAME }}-"UI" -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/ui/resources_windows_arm64.syso
+        run: goversioninfo -arm -64 -icon client/ui/build/windows/icon.ico -manifest client/ui/build/windows/wails.exe.manifest -product-name ${{ env.PRODUCT_NAME }}-"UI" -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/ui/resources_windows_arm64.syso
 
       - name: Run GoReleaser
-        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
+        uses: goreleaser/goreleaser-action@4c6ab561adb47e50c45ef534e2155934e91c40c1 # v7.2.0
         with:
           version: ${{ env.GORELEASER_VER }}
           args: release --config .goreleaser_ui.yaml --clean ${{ env.flags }}
@@ -464,17 +480,17 @@ jobs:
       - if: ${{ !startsWith(github.ref, 'refs/tags/v') }}
         run: echo "flags=--snapshot" >> $GITHUB_ENV
       - name: Checkout
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           fetch-depth: 0 # It is required for GoReleaser to work properly
           persist-credentials: false
       - name: Set up Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
           cache: false
       - name: Cache Go modules
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: |
             ~/go/pkg/mod
@@ -486,9 +502,23 @@ jobs:
         run: go mod tidy
       - name: check git status
         run: git --no-pager diff --exit-code
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+      - name: Set up pnpm
+        uses: pnpm/action-setup@a3252b78c470c02df07e9d59298aecedc3ccdd6d # v3.0.0
+        with:
+          version: 11
+      - name: Install wails3 CLI
+        # Version derived from go.mod so the binding generator always matches
+        # the wails runtime the binary links against.
+        run: |
+          WAILS_VERSION=$(go list -m -f '{{.Version}}' github.com/wailsapp/wails/v3)
+          go install github.com/wailsapp/wails/v3/cmd/wails3@$WAILS_VERSION
       - name: Run GoReleaser
         id: goreleaser
-        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
+        uses: goreleaser/goreleaser-action@4c6ab561adb47e50c45ef534e2155934e91c40c1 # v7.2.0
         with:
           version: ${{ env.GORELEASER_VER }}
           args: release --config .goreleaser_ui_darwin.yaml --clean ${{ env.flags }}
@@ -522,7 +552,7 @@ jobs:
       downloadPath: '${{ github.workspace }}\temp'
     steps:
       - name: Checkout
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
@@ -534,13 +564,13 @@ jobs:
         run: echo "C:\Program Files\7-Zip" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
 
       - name: Download release artifacts
-        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
+        uses: actions/download-artifact@70fc10c6e5e1ce46ad2ea6f2b72d43f7d47b13c3 # v8.0.1
         with:
           name: release
           path: release
 
       - name: Download UI release artifacts
-        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
+        uses: actions/download-artifact@70fc10c6e5e1ce46ad2ea6f2b72d43f7d47b13c3 # v8.0.1
         with:
           name: release-ui
           path: release-ui
@@ -573,23 +603,6 @@ jobs:
       - name: Move wintun.dll into dist
         run: mv ${{ env.downloadPath }}\wintun\bin\${{ matrix.wintun_arch }}\wintun.dll ${{ github.workspace }}\dist\${{ env.PackageWorkdir }}\
 
-      - name: Download Mesa3D (amd64 only)
-        id: download-mesa3d
-        if: matrix.arch == 'amd64'
-        uses: netbirdio/shared-actions/actions/win-download-and-verify@be5df6047383da2236e02243cceb857d8567c27e # v0.0.2
-        with:
-          url: https://pkgs.netbird.io/mesa3d/MesaForWindows-x64-20.1.8.7z
-          destination: ${{ env.downloadPath }}\mesa3d.7z
-          sha256: 71c7cb64ec229a1d6b8d62fa08e1889ed2bd17c0eeede8689daf0f25cb31d6b9
-
-      - name: Extract Mesa3D driver (amd64 only)
-        if: matrix.arch == 'amd64'
-        run: 7z x -o"${{ env.downloadPath }}" "${{ env.downloadPath }}/mesa3d.7z"
-
-      - name: Move opengl32.dll into dist (amd64 only)
-        if: matrix.arch == 'amd64'
-        run: mv ${{ env.downloadPath }}\opengl32.dll ${{ github.workspace }}\dist\${{ env.PackageWorkdir }}\
-
       - name: Download EnVar plugin for NSIS
         uses: netbirdio/shared-actions/actions/win-download-and-verify@be5df6047383da2236e02243cceb857d8567c27e # v0.0.2
         with:
@@ -612,6 +625,28 @@ jobs:
         if: matrix.arch == 'amd64'
         run: 7z x -o"${{ github.workspace }}/NSIS_Plugins" "${{ github.workspace }}/ShellExecAsUser_amd64-Unicode.7z"
 
+      - name: Set up Go for wails3 CLI
+        uses: actions/setup-go@v5
+        with:
+          go-version-file: "go.mod"
+          cache: false
+
+      - name: Install wails3 CLI
+        # Version derived from go.mod so the bootstrapper payload always
+        # matches the wails runtime the binary links against.
+        shell: bash
+        run: |
+          WAILS_VERSION=$(go list -m -f '{{.Version}}' github.com/wailsapp/wails/v3)
+          go install github.com/wailsapp/wails/v3/cmd/wails3@$WAILS_VERSION
+
+      - name: Stage WebView2 bootstrapper for installers
+        # Both client/installer.nsis and client/netbird.wxs reference
+        # client/MicrosoftEdgeWebview2Setup.exe. wails3 writes it there.
+        # The signing pipeline (netbirdio/sign-pipelines) does the same
+        # step for release builds; this mirrors it for PR sanity testing.
+        shell: bash
+        run: wails3 generate webview2bootstrapper -dir client
+
       - name: Build NSIS installer
         shell: pwsh
         env:

.github/workflows/test-infrastructure-files.yml

@@ -68,17 +68,17 @@ jobs:
         run: sudo apt-get install -y curl
 
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
 
       - name: Cache Go modules
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
         with:
           path: ~/go/pkg/mod
           key: ${{ runner.os }}-go-${{ hashFiles('**/go.sum') }}
@@ -207,7 +207,7 @@ jobs:
       - name: Build management docker image
         working-directory: management
         run: |
-          docker build -t netbirdio/management:latest --build-arg TARGETPLATFORM=. .
+          docker build -t netbirdio/management:latest .
 
       - name: Build signal binary
         working-directory: signal
@@ -216,7 +216,7 @@ jobs:
       - name: Build signal docker image
         working-directory: signal
         run: |
-          docker build -t netbirdio/signal:latest --build-arg TARGETPLATFORM=. .
+          docker build -t netbirdio/signal:latest .
 
       - name: Build relay binary
         working-directory: relay
@@ -225,7 +225,7 @@ jobs:
       - name: Build relay docker image
         working-directory: relay
         run: |
-          docker build -t netbirdio/relay:latest --build-arg TARGETPLATFORM=. .
+          docker build -t netbirdio/relay:latest .
 
       - name: run docker compose up
         working-directory: infrastructure_files/artifacts
@@ -256,7 +256,7 @@ jobs:
         run: sudo apt-get install -y jq
 
       - name: Checkout code
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
 

.github/workflows/wasm-build-validation.yml

@@ -19,15 +19,15 @@ jobs:
       GOARCH: wasm
     steps:
       - name: Checkout repository
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
       - name: Install dependencies
-        run: sudo apt update && sudo apt install -y -q libgtk-3-dev libayatana-appindicator3-dev libgl1-mesa-dev xorg-dev libpcap-dev
+        run: sudo apt update && sudo apt install -y -q libgtk-4-dev libwebkitgtk-6.0-dev libsoup-3.0-dev libgl1-mesa-dev xorg-dev libpcap-dev
       - name: Install golangci-lint
         uses: golangci/golangci-lint-action@82606bf257cbaff209d206a39f5134f0cfbfd2ee #v9.2.1
         with:
@@ -44,11 +44,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
         with:
           go-version-file: "go.mod"
       - name: Build Wasm client

.golangci.yaml

@@ -114,6 +114,16 @@ linters:
       - linters:
           - staticcheck
         text: "QF1012"
+      # client/ui/main.go uses //go:embed all:frontend/dist; the
+      # directory is populated by `pnpm build` in the release pipeline
+      # and missing at lint time, so the embed parses to "no matching
+      # files found" — surfaced by golangci-lint's typecheck pre-pass.
+      # Suppress just that one diagnostic; the rest of the package
+      # (services/, tray.go, grpc.go, ...) still gets linted normally.
+      - linters:
+          - typecheck
+        path: client/ui/main\.go
+        text: "pattern all:frontend/dist"
     paths:
       - third_party$
       - builtin$

.goreleaser.yaml

@@ -212,6 +212,7 @@ nfpms:
     description: Netbird client.
     homepage: https://netbird.io/
     license: BSD-3-Clause
+    vendor: NetBird
     id: netbird_deb
     bindir: /usr/bin
     builds:
@@ -226,6 +227,7 @@ nfpms:
     description: Netbird client.
     homepage: https://netbird.io/
     license: BSD-3-Clause
+    vendor: NetBird
     id: netbird_rpm
     bindir: /usr/bin
     builds:
@@ -462,13 +464,9 @@ checksum:
     - glob: ./infrastructure_files/getting-started-with-zitadel.sh
     - glob: ./release_files/install.sh
     - glob: ./infrastructure_files/getting-started.sh
-    - glob: ./infrastructure_files/getting-started-enterprise.sh
-    - glob: ./infrastructure_files/migrate-to-enterprise.sh
 
 release:
   extra_files:
     - glob: ./infrastructure_files/getting-started-with-zitadel.sh
     - glob: ./release_files/install.sh
     - glob: ./infrastructure_files/getting-started.sh
-    - glob: ./infrastructure_files/getting-started-enterprise.sh
-    - glob: ./infrastructure_files/migrate-to-enterprise.sh

.goreleaser_ui.yaml

@@ -2,6 +2,15 @@ version: 2
 env:
   - SKIP_PUBLISH={{ if index .Env "SKIP_PUBLISH" }}{{ .Env.SKIP_PUBLISH }}{{ else }}true{{ end }}
 project_name: netbird-ui
+
+before:
+  hooks:
+    # Bindings are gitignored; regenerate before the frontend build so
+    # the @wailsio/runtime Vite plugin can resolve them (vite refuses to
+    # build without them).
+    - sh -c 'cd client/ui && wails3 generate bindings -clean=true -ts'
+    - sh -c 'cd client/ui/frontend && pnpm install --frozen-lockfile && pnpm build'
+
 builds:
   - id: netbird-ui
     dir: client/ui
@@ -62,6 +71,8 @@ nfpms:
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client UI.
     homepage: https://netbird.io/
+    license: BSD-3-Clause
+    vendor: NetBird
     id: netbird_ui_deb
     package_name: netbird-ui
     builds:
@@ -71,9 +82,9 @@ nfpms:
     scripts:
       postinstall: "release_files/ui-post-install.sh"
     contents:
-      - src: client/ui/build/netbird.desktop
-        dst: /usr/share/applications/netbird.desktop
-      - src: client/ui/assets/netbird.png
+      - src: client/ui/build/linux/netbird.desktop
+        dst: /usr/share/applications/org.wails.netbird.desktop
+      - src: client/ui/build/appicon.png
         dst: /usr/share/pixmaps/netbird.png
     dependencies:
       - netbird
@@ -81,6 +92,8 @@ nfpms:
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client UI.
     homepage: https://netbird.io/
+    license: BSD-3-Clause
+    vendor: NetBird
     id: netbird_ui_rpm
     package_name: netbird-ui
     builds:
@@ -90,12 +103,13 @@ nfpms:
     scripts:
       postinstall: "release_files/ui-post-install.sh"
     contents:
-      - src: client/ui/build/netbird.desktop
-        dst: /usr/share/applications/netbird.desktop
-      - src: client/ui/assets/netbird.png
+      - src: client/ui/build/linux/netbird.desktop
+        dst: /usr/share/applications/org.wails.netbird.desktop
+      - src: client/ui/build/appicon.png
         dst: /usr/share/pixmaps/netbird.png
     dependencies:
       - netbird
+
     rpm:
       signature:
         key_file: '{{ if index .Env "GPG_RPM_KEY_FILE" }}{{ .Env.GPG_RPM_KEY_FILE }}{{ end }}'

.goreleaser_ui_darwin.yaml

@@ -1,6 +1,15 @@
 version: 2
 
 project_name: netbird-ui
+
+before:
+  hooks:
+    # Bindings are gitignored; regenerate before the frontend build so
+    # the @wailsio/runtime Vite plugin can resolve them (vite refuses to
+    # build without them).
+    - sh -c 'cd client/ui && wails3 generate bindings -clean=true -ts'
+    - sh -c 'cd client/ui/frontend && pnpm install --frozen-lockfile && pnpm build'
+
 builds:
   - id: netbird-ui-darwin
     dir: client/ui
@@ -20,8 +29,6 @@ builds:
     ldflags:
       - -s -w -X github.com/netbirdio/netbird/version.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}} -X main.builtBy=goreleaser
     mod_timestamp: "{{ .CommitTimestamp }}"
-    tags:
-      - load_wgnt_from_rsrc
 
 universal_binaries:
   - id: netbird-ui-darwin

CONTRIBUTING.md

@@ -79,13 +79,21 @@ dependencies are installed. Here is a short guide on how that can be done.
 
 ### Requirements
 
-#### Go 1.21
+#### Go 1.25
 
 Follow the installation guide from https://go.dev/
 
-#### UI client - Fyne toolkit 
+#### UI client - Wails v3 + React
 
-We use the fyne toolkit in our UI client. You can follow its requirement guide to have all its dependencies installed: https://developer.fyne.io/started/#prerequisites
+The desktop UI client (`client/ui`) is built with [Wails v3](https://v3.wails.io/) and a React frontend rendered in a WebView. To build it you need:
+
+- Go ≥ 1.25
+- Node ≥ 20 and **pnpm** (`corepack enable && corepack prepare pnpm@latest --activate`)
+- The `wails3` CLI: `go install github.com/wailsapp/wails/v3/cmd/wails3@latest`
+- The `task` runner: `go install github.com/go-task/task/v3/cmd/task@latest`
+- Linux only: `libwebkitgtk-6.0-dev`, `libgtk-4-dev`, `libsoup-3.0-dev`
+
+All UI build, dev-loop, and cross-compile commands are described in the [UI client](#ui-client) section below.
 
 #### gRPC
 You can follow the instructions from the quickstarter guide https://grpc.io/docs/languages/go/quickstart/#prerequisites and then run the `generate.sh` files located in each `proto` directory to generate changes.
@@ -214,6 +222,39 @@ To start NetBird the client in the foreground:
 sudo ./client up --log-level debug --log-file console
 ```
 > On Windows use a powershell with administrator privileges
+
+#### UI client
+
+The desktop UI lives in `client/ui` and is built with Wails v3 (see [Requirements](#ui-client---wails-v3--react)). All commands run from `client/ui`.
+
+Live-reload development (Vite + Go binary + `*.go` watcher):
+
+```
+cd client/ui
+task dev
+```
+
+Pass daemon flags after `--`:
+
+```
+task dev -- --daemon-addr=tcp://127.0.0.1:41731
+```
+
+Production build (frontend assets embedded into the binary, output in `client/ui/bin/`):
+
+```
+cd client/ui
+task build
+```
+
+Cross-compile the Windows binary from Linux (requires the mingw-w64 toolchain, e.g. `sudo apt install gcc-mingw-w64-x86-64`):
+
+```
+CGO_ENABLED=1 task windows:build
+```
+
+> macOS cross-compile from Linux is not supported (signing and notarization need a real Mac).
+
 #### Signal service
 
 To start NetBird's signal, execute:
@@ -251,10 +292,10 @@ Create dist directory
 mkdir -p dist/netbird_windows_amd64
 ```
 
-UI client
+UI client (built with Wails v3 — see the [UI client](#ui-client) section above)
 ```shell
-CC=x86_64-w64-mingw32-gcc CGO_ENABLED=1 GOOS=windows GOARCH=amd64 go build -o netbird-ui.exe -ldflags "-s -w -H windowsgui" ./client/ui
-mv netbird-ui.exe ./dist/netbird_windows_amd64/
+(cd client/ui && CGO_ENABLED=1 task windows:build)
+mv client/ui/bin/netbird-ui.exe ./dist/netbird_windows_amd64/
 ```
 
 Client
@@ -291,8 +332,6 @@ go test -exec sudo ./...
 ```
 > On Windows use a powershell with administrator privileges
 
-> Non-GTK environments will need the `libayatana-appindicator3-dev` (debian/ubuntu) package installed
-
 ## Checklist before submitting a PR
 As a critical network service and open-source project, we must enforce a few things before submitting the pull-requests:
 - Keep functions as simple as possible, with a single purpose

Makefile

@@ -1,4 +1,4 @@
-.PHONY: lint lint-all lint-install setup-hooks test-unit test-privileged
+.PHONY: lint lint-all lint-install setup-hooks
 GOLANGCI_LINT := $(shell pwd)/bin/golangci-lint
 
 # Install golangci-lint locally if needed
@@ -25,15 +25,3 @@ setup-hooks:
 	@git config core.hooksPath .githooks
 	@chmod +x .githooks/pre-push
 	@echo "✅ Git hooks configured! Pre-push will now run 'make lint'"
-
-# Host-safe unit tests: excludes the privileged-tagged tests (root / system-mutating).
-# Runs as a normal user with no sudo and leaves host networking untouched.
-test-unit:
-	@go test -tags devcert -timeout 10m ./...
-
-# Privileged suite: runs the `privileged`-tagged tests inside a --privileged
-# --cap-add=NET_ADMIN container via the ory/dockertest harness. Requires Docker.
-# Narrow the run with env vars, e.g.:
-#   PRIV_RUN=TestNftablesManager PRIV_PKGS=./client/firewall/nftables/... make test-privileged
-test-privileged:
-	@go test -tags 'devcert privileged' -timeout 30m -run TestRunPrivilegedSuiteInDocker -v ./client/testutil/privileged/...

README.md

@@ -33,15 +33,10 @@
   <br/>
   <br/>
   <strong>
-    🚀 <a href="https://netbird.io/careers">We are hiring! Join us at https://netbird.io/careers</a>
+    🚀 <a href="https://careers.netbird.io">We are hiring! Join us at careers.netbird.io</a>
   </strong>
 </p>
 
-> ### 🤖 NetBird Agent Network (Beta)
-> Identity-aware access control for AI agents — keyless access to LLM APIs and private
-> resources over the encrypted NetBird tunnel. See [`agent-network/`](agent-network/) or
-> read the docs at **[netbird.ai](https://netbird.ai)**.
-
 **NetBird combines a configuration-free peer-to-peer private network and a centralized access control system in a single platform, making it easy to create secure private networks for your organization or home.**
 
 **Connect.** NetBird creates a WireGuard-based overlay network that automatically connects your machines over an encrypted tunnel, leaving behind the hassle of opening ports, complex firewall rules, VPN gateways, and so forth.

agent-network/README.md

@@ -1,39 +0,0 @@
-# NetBird Agent Network
-
-Agent Network is NetBird's access control layer for AI agents and the people who run
-them. It gives every agent a real identity, tied to your identity provider (IdP), and
-governs what it can reach — the LLM APIs and AI gateways it can call, and the internal
-resources it can access. Traffic flows only over the encrypted NetBird tunnel, scoped by
-policy, with no API keys to leak.
-
-> **Beta.** Agent Network is open source and can be self-hosted on your own
-> infrastructure.
-
-## How it works
-
-Agent Network is built on two existing NetBird capabilities:
-
-- **Overlay network** — the encrypted WireGuard mesh between peers.
-- **Reverse proxy** — a NetBird peer that terminates LLM requests, establishes the
-  caller's identity, evaluates policies/limits/guardrails, injects the upstream provider
-  key server-side, forwards to the API or gateway, and records usage.
-
-LLM traffic is routed through the proxy's identity-aware pipeline, while internal
-resources (databases, internal APIs, self-hosted models) are reached directly over
-peer-to-peer WireGuard tunnels, governed by the same identities and access policies.
-
-## Where the code lives
-
-There is no separate "agent-network" service — it reuses the reverse-proxy and management
-components:
-
-- [`proxy/`](../proxy) — the NetBird reverse proxy that serves the agent network endpoint
-  and runs the per-request middleware pipeline.
-- [`management/internals/modules/reverseproxy/`](../management/internals/modules/reverseproxy)
-  — the management-side control plane: providers, policies, guardrails, limits, routing,
-  and usage/access logs.
-
-## Documentation
-
-Full documentation, architecture, and quickstart:
-**https://docs.netbird.io/agent-network**

client/cmd/debug.go

@@ -130,7 +130,7 @@ func debugConfigDump(cmd *cobra.Command, _ []string) error {
 
 	client := proto.NewDaemonServiceClient(conn)
 	resp, err := client.GetConfig(cmd.Context(), &proto.GetConfigRequest{
-		ProfileName: string(activeProf.ID),
+		ProfileName: activeProf.Name,
 		Username:    currUser.Username,
 	})
 	if err != nil {

client/cmd/login.go

@@ -22,11 +22,19 @@ import (
 	"github.com/netbirdio/netbird/util"
 )
 
+// extendSessionFlag drives the `netbird login --extend` flow: refresh the
+// SSO session expiry on the management server without tearing down the
+// tunnel. Mutually exclusive with setup-key login (a setup-key cannot
+// refresh an SSO-tracked peer — see auth.errSetupKeyOnSSOExpiredPeer).
+var extendSessionFlag bool
+
 func init() {
 	loginCmd.PersistentFlags().BoolVar(&noBrowser, noBrowserFlag, false, noBrowserDesc)
 	loginCmd.PersistentFlags().BoolVar(&showQR, showQRFlag, false, showQRDesc)
 	loginCmd.PersistentFlags().StringVar(&profileName, profileNameFlag, "", profileNameDesc)
 	loginCmd.PersistentFlags().StringVarP(&configPath, "config", "c", "", "(DEPRECATED) Netbird config file location")
+	loginCmd.PersistentFlags().BoolVar(&extendSessionFlag, "extend", false,
+		"refresh the SSO session expiry without tearing down the tunnel (requires an active connection)")
 }
 
 var loginCmd = &cobra.Command{
@@ -61,6 +69,16 @@ var loginCmd = &cobra.Command{
 			return err
 		}
 
+		if extendSessionFlag {
+			if providedSetupKey != "" {
+				return fmt.Errorf("--extend cannot be combined with a setup key; setup keys can only enrol new peers")
+			}
+			if err := doExtendSession(ctx, cmd); err != nil {
+				return fmt.Errorf("extend session failed: %v", err)
+			}
+			return nil
+		}
+
 		// workaround to run without service
 		if util.FindFirstLogPath(logFiles) == "" {
 			if err := doForegroundLogin(ctx, cmd, providedSetupKey, activeProf); err != nil {
@@ -152,6 +170,65 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
 	return nil
 }
 
+// doExtendSession drives the daemon's RequestExtendAuthSession /
+// WaitExtendAuthSession pair. The user is sent through a regular SSO flow
+// (browser + verification URL) and the resulting JWT is forwarded to the
+// management server's ExtendAuthSession RPC. The tunnel stays up
+// throughout — no Down/Up, no network-map resync.
+func doExtendSession(ctx context.Context, cmd *cobra.Command) error {
+	conn, err := DialClientGRPCServer(ctx, daemonAddr)
+	if err != nil {
+		//nolint
+		return fmt.Errorf("failed to connect to daemon error: %v\n"+
+			"If the daemon is not running please run: "+
+			"\nnetbird service install \nnetbird service start\n", err)
+	}
+	defer conn.Close()
+
+	client := proto.NewDaemonServiceClient(conn)
+
+	req := &proto.RequestExtendAuthSessionRequest{}
+	// Pre-fill the IdP login hint from the active profile so the user
+	// doesn't have to retype their email. Best-effort: we still proceed
+	// without a hint if the lookup fails.
+	pm := profilemanager.NewProfileManager()
+	if active, perr := pm.GetActiveProfile(); perr == nil {
+		if profState, sperr := pm.GetProfileState(active.ID); sperr == nil && profState.Email != "" {
+			req.Hint = &profState.Email
+		}
+	}
+
+	startResp, err := client.RequestExtendAuthSession(ctx, req)
+	if err != nil {
+		return fmt.Errorf("start extend session: %v", err)
+	}
+
+	uri := startResp.GetVerificationURIComplete()
+	if uri == "" {
+		uri = startResp.GetVerificationURI()
+	}
+	openURL(cmd, uri, startResp.GetUserCode(), noBrowser, showQR)
+
+	waitResp, err := client.WaitExtendAuthSession(ctx, &proto.WaitExtendAuthSessionRequest{
+		DeviceCode: startResp.GetDeviceCode(),
+		UserCode:   startResp.GetUserCode(),
+	})
+	if err != nil {
+		return fmt.Errorf("wait for extend session: %v", err)
+	}
+
+	if ts := waitResp.GetSessionExpiresAt(); ts.IsValid() && !ts.AsTime().IsZero() {
+		deadline := ts.AsTime().Local()
+		cmd.Printf("Session extended. New expiry: %s\n", deadline.Format("2006-01-02 15:04:05 MST"))
+	} else {
+		// Management reported the peer is not eligible (e.g. login
+		// expiration disabled on the account). Surface that fact
+		// instead of pretending the call succeeded.
+		cmd.Println("Session extension call completed, but the management server did not return a new deadline (peer may not be SSO-tracked or login expiration is disabled).")
+	}
+	return nil
+}
+
 func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, profileName string, username string) (*profilemanager.Profile, error) {
 	// switch profile if provided
 

client/cmd/service_privileged_test.go

@@ -1,196 +0,0 @@
-//go:build privileged
-
-package cmd
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"runtime"
-	"testing"
-	"time"
-
-	"github.com/kardianos/service"
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-const (
-	serviceStartTimeout = 10 * time.Second
-	serviceStopTimeout  = 5 * time.Second
-	statusPollInterval  = 500 * time.Millisecond
-)
-
-// waitForServiceStatus waits for service to reach expected status with timeout
-func waitForServiceStatus(expectedStatus service.Status, timeout time.Duration) (bool, error) {
-	cfg, err := newSVCConfig()
-	if err != nil {
-		return false, err
-	}
-
-	ctxSvc, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-	if err != nil {
-		return false, err
-	}
-
-	ctx, timeoutCancel := context.WithTimeout(context.Background(), timeout)
-	defer timeoutCancel()
-
-	ticker := time.NewTicker(statusPollInterval)
-	defer ticker.Stop()
-
-	for {
-		select {
-		case <-ctx.Done():
-			return false, fmt.Errorf("timeout waiting for service status %v", expectedStatus)
-		case <-ticker.C:
-			status, err := s.Status()
-			if err != nil {
-				// Continue polling on transient errors
-				continue
-			}
-			if status == expectedStatus {
-				return true, nil
-			}
-		}
-	}
-}
-
-// TestServiceLifecycle tests the complete service lifecycle
-func TestServiceLifecycle(t *testing.T) {
-	// TODO: Add support for Windows and macOS
-	if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
-		t.Skipf("Skipping service lifecycle test on unsupported OS: %s", runtime.GOOS)
-	}
-
-	if os.Getenv("CONTAINER") == "true" {
-		t.Skip("Skipping service lifecycle test in container environment")
-	}
-
-	originalServiceName := serviceName
-	serviceName = "netbirdtest" + fmt.Sprintf("%d", time.Now().Unix())
-	defer func() {
-		serviceName = originalServiceName
-	}()
-
-	tempDir := t.TempDir()
-	configPath = fmt.Sprintf("%s/netbird-test-config.json", tempDir)
-	logLevel = "info"
-	daemonAddr = fmt.Sprintf("unix://%s/netbird-test.sock", tempDir)
-
-	// Ensure cleanup even if a subtest fails and Stop/Uninstall subtests don't run.
-	t.Cleanup(func() {
-		cfg, err := newSVCConfig()
-		if err != nil {
-			t.Errorf("cleanup: create service config: %v", err)
-			return
-		}
-		ctxSvc, cancel := context.WithCancel(context.Background())
-		defer cancel()
-		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-		if err != nil {
-			t.Errorf("cleanup: create service: %v", err)
-			return
-		}
-
-		// If the subtests already cleaned up, there's nothing to do.
-		if _, err := s.Status(); err != nil {
-			return
-		}
-
-		if err := s.Stop(); err != nil {
-			t.Errorf("cleanup: stop service: %v", err)
-		}
-		if err := s.Uninstall(); err != nil {
-			t.Errorf("cleanup: uninstall service: %v", err)
-		}
-	})
-
-	ctx := context.Background()
-
-	t.Run("Install", func(t *testing.T) {
-		installCmd.SetContext(ctx)
-		err := installCmd.RunE(installCmd, []string{})
-		require.NoError(t, err)
-
-		cfg, err := newSVCConfig()
-		require.NoError(t, err)
-
-		ctxSvc, cancel := context.WithCancel(context.Background())
-		defer cancel()
-
-		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-		require.NoError(t, err)
-
-		status, err := s.Status()
-		assert.NoError(t, err)
-		assert.NotEqual(t, service.StatusUnknown, status)
-	})
-
-	t.Run("Start", func(t *testing.T) {
-		startCmd.SetContext(ctx)
-		err := startCmd.RunE(startCmd, []string{})
-		require.NoError(t, err)
-
-		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
-		require.NoError(t, err)
-		assert.True(t, running)
-	})
-
-	t.Run("Restart", func(t *testing.T) {
-		restartCmd.SetContext(ctx)
-		err := restartCmd.RunE(restartCmd, []string{})
-		require.NoError(t, err)
-
-		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
-		require.NoError(t, err)
-		assert.True(t, running)
-	})
-
-	t.Run("Reconfigure", func(t *testing.T) {
-		originalLogLevel := logLevel
-		logLevel = "debug"
-		defer func() {
-			logLevel = originalLogLevel
-		}()
-
-		reconfigureCmd.SetContext(ctx)
-		err := reconfigureCmd.RunE(reconfigureCmd, []string{})
-		require.NoError(t, err)
-
-		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
-		require.NoError(t, err)
-		assert.True(t, running)
-	})
-
-	t.Run("Stop", func(t *testing.T) {
-		stopCmd.SetContext(ctx)
-		err := stopCmd.RunE(stopCmd, []string{})
-		require.NoError(t, err)
-
-		stopped, err := waitForServiceStatus(service.StatusStopped, serviceStopTimeout)
-		require.NoError(t, err)
-		assert.True(t, stopped)
-	})
-
-	t.Run("Uninstall", func(t *testing.T) {
-		uninstallCmd.SetContext(ctx)
-		err := uninstallCmd.RunE(uninstallCmd, []string{})
-		require.NoError(t, err)
-
-		cfg, err := newSVCConfig()
-		require.NoError(t, err)
-
-		ctxSvc, cancel := context.WithCancel(context.Background())
-		defer cancel()
-
-		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-		require.NoError(t, err)
-
-		_, err = s.Status()
-		assert.Error(t, err)
-	})
-}

client/cmd/service_test.go

@@ -1,12 +1,16 @@
 package cmd
 
 import (
+	"context"
+	"fmt"
 	"os"
 	"os/signal"
 	"runtime"
 	"syscall"
 	"testing"
+	"time"
 
+	"github.com/kardianos/service"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -27,6 +31,186 @@ func TestMain(m *testing.M) {
 	os.Exit(m.Run())
 }
 
+const (
+	serviceStartTimeout = 10 * time.Second
+	serviceStopTimeout  = 5 * time.Second
+	statusPollInterval  = 500 * time.Millisecond
+)
+
+// waitForServiceStatus waits for service to reach expected status with timeout
+func waitForServiceStatus(expectedStatus service.Status, timeout time.Duration) (bool, error) {
+	cfg, err := newSVCConfig()
+	if err != nil {
+		return false, err
+	}
+
+	ctxSvc, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+	if err != nil {
+		return false, err
+	}
+
+	ctx, timeoutCancel := context.WithTimeout(context.Background(), timeout)
+	defer timeoutCancel()
+
+	ticker := time.NewTicker(statusPollInterval)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return false, fmt.Errorf("timeout waiting for service status %v", expectedStatus)
+		case <-ticker.C:
+			status, err := s.Status()
+			if err != nil {
+				// Continue polling on transient errors
+				continue
+			}
+			if status == expectedStatus {
+				return true, nil
+			}
+		}
+	}
+}
+
+// TestServiceLifecycle tests the complete service lifecycle
+func TestServiceLifecycle(t *testing.T) {
+	// TODO: Add support for Windows and macOS
+	if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
+		t.Skipf("Skipping service lifecycle test on unsupported OS: %s", runtime.GOOS)
+	}
+
+	if os.Getenv("CONTAINER") == "true" {
+		t.Skip("Skipping service lifecycle test in container environment")
+	}
+
+	originalServiceName := serviceName
+	serviceName = "netbirdtest" + fmt.Sprintf("%d", time.Now().Unix())
+	defer func() {
+		serviceName = originalServiceName
+	}()
+
+	tempDir := t.TempDir()
+	configPath = fmt.Sprintf("%s/netbird-test-config.json", tempDir)
+	logLevel = "info"
+	daemonAddr = fmt.Sprintf("unix://%s/netbird-test.sock", tempDir)
+
+	// Ensure cleanup even if a subtest fails and Stop/Uninstall subtests don't run.
+	t.Cleanup(func() {
+		cfg, err := newSVCConfig()
+		if err != nil {
+			t.Errorf("cleanup: create service config: %v", err)
+			return
+		}
+		ctxSvc, cancel := context.WithCancel(context.Background())
+		defer cancel()
+		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+		if err != nil {
+			t.Errorf("cleanup: create service: %v", err)
+			return
+		}
+
+		// If the subtests already cleaned up, there's nothing to do.
+		if _, err := s.Status(); err != nil {
+			return
+		}
+
+		if err := s.Stop(); err != nil {
+			t.Errorf("cleanup: stop service: %v", err)
+		}
+		if err := s.Uninstall(); err != nil {
+			t.Errorf("cleanup: uninstall service: %v", err)
+		}
+	})
+
+	ctx := context.Background()
+
+	t.Run("Install", func(t *testing.T) {
+		installCmd.SetContext(ctx)
+		err := installCmd.RunE(installCmd, []string{})
+		require.NoError(t, err)
+
+		cfg, err := newSVCConfig()
+		require.NoError(t, err)
+
+		ctxSvc, cancel := context.WithCancel(context.Background())
+		defer cancel()
+
+		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+		require.NoError(t, err)
+
+		status, err := s.Status()
+		assert.NoError(t, err)
+		assert.NotEqual(t, service.StatusUnknown, status)
+	})
+
+	t.Run("Start", func(t *testing.T) {
+		startCmd.SetContext(ctx)
+		err := startCmd.RunE(startCmd, []string{})
+		require.NoError(t, err)
+
+		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
+		require.NoError(t, err)
+		assert.True(t, running)
+	})
+
+	t.Run("Restart", func(t *testing.T) {
+		restartCmd.SetContext(ctx)
+		err := restartCmd.RunE(restartCmd, []string{})
+		require.NoError(t, err)
+
+		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
+		require.NoError(t, err)
+		assert.True(t, running)
+	})
+
+	t.Run("Reconfigure", func(t *testing.T) {
+		originalLogLevel := logLevel
+		logLevel = "debug"
+		defer func() {
+			logLevel = originalLogLevel
+		}()
+
+		reconfigureCmd.SetContext(ctx)
+		err := reconfigureCmd.RunE(reconfigureCmd, []string{})
+		require.NoError(t, err)
+
+		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
+		require.NoError(t, err)
+		assert.True(t, running)
+	})
+
+	t.Run("Stop", func(t *testing.T) {
+		stopCmd.SetContext(ctx)
+		err := stopCmd.RunE(stopCmd, []string{})
+		require.NoError(t, err)
+
+		stopped, err := waitForServiceStatus(service.StatusStopped, serviceStopTimeout)
+		require.NoError(t, err)
+		assert.True(t, stopped)
+	})
+
+	t.Run("Uninstall", func(t *testing.T) {
+		uninstallCmd.SetContext(ctx)
+		err := uninstallCmd.RunE(uninstallCmd, []string{})
+		require.NoError(t, err)
+
+		cfg, err := newSVCConfig()
+		require.NoError(t, err)
+
+		ctxSvc, cancel := context.WithCancel(context.Background())
+		defer cancel()
+
+		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+		require.NoError(t, err)
+
+		_, err = s.Status()
+		assert.Error(t, err)
+	})
+}
+
 // TestServiceEnvVars tests environment variable parsing
 func TestServiceEnvVars(t *testing.T) {
 	tests := []struct {

client/cmd/status.go

@@ -6,6 +6,7 @@ import (
 	"net"
 	"net/netip"
 	"strings"
+	"time"
 
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/status"
@@ -115,6 +116,11 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 	// manager only knows the active profile ID, not its display name.
 	profName := getActiveProfileName(ctx)
 
+	var sessionExpiresAt time.Time
+	if ts := resp.GetSessionExpiresAt(); ts.IsValid() {
+		sessionExpiresAt = ts.AsTime().UTC()
+	}
+
 	var outputInformationHolder = nbstatus.ConvertToStatusOutputOverview(resp.GetFullStatus(), nbstatus.ConvertOptions{
 		Anonymize:            anonymizeFlag,
 		DaemonVersion:        resp.GetDaemonVersion(),
@@ -125,6 +131,7 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 		IPsFilter:            ipsFilterMap,
 		ConnectionTypeFilter: connectionTypeFilter,
 		ProfileName:          profName,
+		SessionExpiresAt:     sessionExpiresAt,
 	})
 	var statusOutputString string
 	switch {

client/embed/embed.go

@@ -279,11 +279,9 @@ func (c *Client) Start(startCtx context.Context) error {
 
 	select {
 	case <-startCtx.Done():
-		// ConnectClient.Stop now cancels its own run context and waits for the
-		// run loop to tear the engine down, so this cancel() is no longer
-		// required to break the deadlock and could be removed. It is kept as a
-		// defensive belt-and-suspenders: cancelling the parent context first
-		// guarantees the run loop is unblocked even if Stop's contract regresses.
+		// Cancel the client context before stopping: Engine.Start blocks on the
+		// signal stream while holding the engine mutex and only unblocks on
+		// cancellation. Stopping first would deadlock on that mutex.
 		cancel()
 		if stopErr := client.Stop(); stopErr != nil {
 			return fmt.Errorf("stop error after context done. Stop error: %w. Context done: %w", stopErr, startCtx.Err())
@@ -470,7 +468,7 @@ func (c *Client) Status() (peer.FullStatus, error) {
 	if connect != nil {
 		engine := connect.Engine()
 		if engine != nil {
-			_ = engine.RunHealthProbes(false)
+			_ = engine.RunHealthProbes(context.Background(), false)
 		}
 	}
 

client/firewall/iptables/manager_linux_test.go

@@ -1,5 +1,3 @@
-//go:build privileged
-
 package iptables
 
 import (

client/firewall/iptables/router_linux_test.go

@@ -1,4 +1,4 @@
-//go:build !android && privileged
+//go:build !android
 
 package iptables
 

client/firewall/nftables/manager_linux_test.go

@@ -1,5 +1,3 @@
-//go:build privileged
-
 package nftables
 
 import (

client/firewall/nftables/router_linux_test.go

@@ -1,4 +1,4 @@
-//go:build !android && privileged
+//go:build !android
 
 package nftables
 

client/iface/iface_test.go

@@ -1,5 +1,3 @@
-//go:build privileged
-
 package iface
 
 import (

client/iface/wgproxy/bind/proxy.go

@@ -136,11 +136,6 @@ func (p *ProxyBind) CloseConn() error {
 	return p.close()
 }
 
-// InjectPacket is a no-op for the userspace proxy: first-packet reinjection is kernel-only.
-func (p *ProxyBind) InjectPacket(_ []byte) error {
-	return nil
-}
-
 func (p *ProxyBind) close() error {
 	if p.remoteConn == nil {
 		return nil

client/iface/wgproxy/ebpf/wrapper.go

@@ -219,17 +219,6 @@ func (p *ProxyWrapper) RedirectAs(endpoint *net.UDPAddr) {
 	p.pausedCond.L.Unlock()
 }
 
-// InjectPacket writes b to the remote peer over the underlying transport.
-func (p *ProxyWrapper) InjectPacket(b []byte) error {
-	if p.remoteConn == nil {
-		return errors.New("proxy not started")
-	}
-	if _, err := p.remoteConn.Write(b); err != nil {
-		return err
-	}
-	return nil
-}
-
 // CloseConn close the remoteConn and automatically remove the conn instance from the map
 func (p *ProxyWrapper) CloseConn() error {
 	if p.cancel == nil {

client/iface/wgproxy/proxy.go

@@ -18,9 +18,4 @@ type Proxy interface {
 	RedirectAs(endpoint *net.UDPAddr)
 	CloseConn() error
 	SetDisconnectListener(disconnected func())
-
-	// InjectPacket writes a raw packet directly to the remote peer over the underlying transport,
-	// bypassing WireGuard. Used to replay the captured lazyconn handshake initiation. Only the
-	// kernel-mode proxies act on it; the userspace proxy is a no-op since reinjection is kernel-only.
-	InjectPacket(b []byte) error
 }

client/iface/wgproxy/proxy_linux_test.go

@@ -1,4 +1,4 @@
-//go:build linux && !android && privileged
+//go:build linux && !android
 
 package wgproxy
 

client/iface/wgproxy/proxy_seed_test.go

@@ -1,4 +1,4 @@
-//go:build !linux || !privileged
+//go:build !linux
 
 package wgproxy
 

client/iface/wgproxy/redirect_test.go

@@ -1,4 +1,4 @@
-//go:build linux && !android && privileged
+//go:build linux && !android
 
 package wgproxy
 
@@ -26,6 +26,64 @@ func compareUDPAddr(addr1, addr2 net.Addr) bool {
 	return udpAddr1.IP.Equal(udpAddr2.IP) && udpAddr1.Port == udpAddr2.Port
 }
 
+// TestRedirectAs_eBPF_IPv4 tests RedirectAs with eBPF proxy using IPv4 addresses
+func TestRedirectAs_eBPF_IPv4(t *testing.T) {
+	wgPort := 51850
+	ebpfProxy := ebpf.NewWGEBPFProxy(wgPort, 1280)
+	if err := ebpfProxy.Listen(); err != nil {
+		t.Fatalf("failed to initialize ebpf proxy: %v", err)
+	}
+	defer func() {
+		if err := ebpfProxy.Free(); err != nil {
+			t.Errorf("failed to free ebpf proxy: %v", err)
+		}
+	}()
+
+	proxy := ebpf.NewProxyWrapper(ebpfProxy)
+
+	// NetBird UDP address of the remote peer
+	nbAddr := &net.UDPAddr{
+		IP:   net.ParseIP("100.108.111.177"),
+		Port: 38746,
+	}
+
+	p2pEndpoint := &net.UDPAddr{
+		IP:   net.ParseIP("192.168.0.56"),
+		Port: 51820,
+	}
+
+	testRedirectAs(t, proxy, wgPort, nbAddr, p2pEndpoint)
+}
+
+// TestRedirectAs_eBPF_IPv6 tests RedirectAs with eBPF proxy using IPv6 addresses
+func TestRedirectAs_eBPF_IPv6(t *testing.T) {
+	wgPort := 51851
+	ebpfProxy := ebpf.NewWGEBPFProxy(wgPort, 1280)
+	if err := ebpfProxy.Listen(); err != nil {
+		t.Fatalf("failed to initialize ebpf proxy: %v", err)
+	}
+	defer func() {
+		if err := ebpfProxy.Free(); err != nil {
+			t.Errorf("failed to free ebpf proxy: %v", err)
+		}
+	}()
+
+	proxy := ebpf.NewProxyWrapper(ebpfProxy)
+
+	// NetBird UDP address of the remote peer
+	nbAddr := &net.UDPAddr{
+		IP:   net.ParseIP("100.108.111.177"),
+		Port: 38746,
+	}
+
+	p2pEndpoint := &net.UDPAddr{
+		IP:   net.ParseIP("fe80::56"),
+		Port: 51820,
+	}
+
+	testRedirectAs(t, proxy, wgPort, nbAddr, p2pEndpoint)
+}
+
 // TestRedirectAs_UDP_IPv4 tests RedirectAs with UDP proxy using IPv4 addresses
 func TestRedirectAs_UDP_IPv4(t *testing.T) {
 	wgPort := 51852
@@ -198,64 +256,6 @@ func testRedirectAs(t *testing.T, proxy Proxy, wgPort int, nbAddr, p2pEndpoint *
 	}
 }
 
-// TestRedirectAs_eBPF_IPv4 tests RedirectAs with eBPF proxy using IPv4 addresses
-func TestRedirectAs_eBPF_IPv4(t *testing.T) {
-	wgPort := 51850
-	ebpfProxy := ebpf.NewWGEBPFProxy(wgPort, 1280)
-	if err := ebpfProxy.Listen(); err != nil {
-		t.Fatalf("failed to initialize ebpf proxy: %v", err)
-	}
-	defer func() {
-		if err := ebpfProxy.Free(); err != nil {
-			t.Errorf("failed to free ebpf proxy: %v", err)
-		}
-	}()
-
-	proxy := ebpf.NewProxyWrapper(ebpfProxy)
-
-	// NetBird UDP address of the remote peer
-	nbAddr := &net.UDPAddr{
-		IP:   net.ParseIP("100.108.111.177"),
-		Port: 38746,
-	}
-
-	p2pEndpoint := &net.UDPAddr{
-		IP:   net.ParseIP("192.168.0.56"),
-		Port: 51820,
-	}
-
-	testRedirectAs(t, proxy, wgPort, nbAddr, p2pEndpoint)
-}
-
-// TestRedirectAs_eBPF_IPv6 tests RedirectAs with eBPF proxy using IPv6 addresses
-func TestRedirectAs_eBPF_IPv6(t *testing.T) {
-	wgPort := 51851
-	ebpfProxy := ebpf.NewWGEBPFProxy(wgPort, 1280)
-	if err := ebpfProxy.Listen(); err != nil {
-		t.Fatalf("failed to initialize ebpf proxy: %v", err)
-	}
-	defer func() {
-		if err := ebpfProxy.Free(); err != nil {
-			t.Errorf("failed to free ebpf proxy: %v", err)
-		}
-	}()
-
-	proxy := ebpf.NewProxyWrapper(ebpfProxy)
-
-	// NetBird UDP address of the remote peer
-	nbAddr := &net.UDPAddr{
-		IP:   net.ParseIP("100.108.111.177"),
-		Port: 38746,
-	}
-
-	p2pEndpoint := &net.UDPAddr{
-		IP:   net.ParseIP("fe80::56"),
-		Port: 51820,
-	}
-
-	testRedirectAs(t, proxy, wgPort, nbAddr, p2pEndpoint)
-}
-
 // TestRedirectAs_Multiple_Switches tests switching between multiple endpoints
 func TestRedirectAs_Multiple_Switches(t *testing.T) {
 	wgPort := 51856

client/iface/wgproxy/udp/proxy.go

@@ -147,17 +147,6 @@ func (p *WGUDPProxy) RedirectAs(endpoint *net.UDPAddr) {
 	p.sendPkg = p.srcFakerConn.SendPkg
 }
 
-// InjectPacket writes b to the remote peer over the underlying transport.
-func (p *WGUDPProxy) InjectPacket(b []byte) error {
-	if p.remoteConn == nil {
-		return errors.New("proxy not started")
-	}
-	if _, err := p.remoteConn.Write(b); err != nil {
-		return err
-	}
-	return nil
-}
-
 // CloseConn close the localConn
 func (p *WGUDPProxy) CloseConn() error {
 	if p.cancel == nil {

client/installer.nsis

@@ -6,7 +6,7 @@
 !define DESCRIPTION "Connect your devices into a secure WireGuard-based overlay network with SSO, MFA, and granular access controls."
 !define INSTALLER_NAME "netbird-installer.exe"
 !define MAIN_APP_EXE "Netbird"
-!define ICON "ui\\assets\\netbird.ico"
+!define ICON "ui\\build\\windows\\icon.ico"
 !define BANNER "ui\\build\\banner.bmp"
 !define LICENSE_DATA "..\\LICENSE"
 
@@ -280,6 +280,43 @@ CreateShortCut "$SMPROGRAMS\${APP_NAME}.lnk" "$INSTDIR\${UI_APP_EXE}"
 CreateShortCut "$DESKTOP\${APP_NAME}.lnk" "$INSTDIR\${UI_APP_EXE}"
 SectionEnd
 
+# Install the Microsoft Edge WebView2 runtime if it isn't already present.
+# Macro adapted from Wails3's NSIS template (wails_tools.nsh): a registry
+# probe followed by a silent install of the embedded evergreen bootstrapper.
+# The MicrosoftEdgeWebview2Setup.exe payload is staged next to this script
+# by the sign-pipelines build step (`wails3 generate webview2bootstrapper`).
+!macro nb.webview2runtime
+    SetRegView 64
+    # Per-machine install marker — populated when the runtime ships with
+    # Edge or has been installed by an admin previously.
+    ReadRegStr $0 HKLM "SOFTWARE\WOW6432Node\Microsoft\EdgeUpdate\Clients\{F3017226-FE2A-4295-8BDF-00C3A9A7E4C5}" "pv"
+    ${If} $0 != ""
+        Goto webview2_ok
+    ${EndIf}
+    # Per-user fallback for HKCU installs.
+    ReadRegStr $0 HKCU "Software\Microsoft\EdgeUpdate\Clients\{F3017226-FE2A-4295-8BDF-00C3A9A7E4C5}" "pv"
+    ${If} $0 != ""
+        Goto webview2_ok
+    ${EndIf}
+
+    SetDetailsPrint both
+    DetailPrint "Installing: WebView2 Runtime"
+    SetDetailsPrint listonly
+
+    InitPluginsDir
+    CreateDirectory "$pluginsdir\webview2bootstrapper"
+    SetOutPath "$pluginsdir\webview2bootstrapper"
+    File "MicrosoftEdgeWebview2Setup.exe"
+    ExecWait '"$pluginsdir\webview2bootstrapper\MicrosoftEdgeWebview2Setup.exe" /silent /install'
+
+    SetDetailsPrint both
+    webview2_ok:
+!macroend
+
+Section -WebView2
+    !insertmacro nb.webview2runtime
+SectionEnd
+
 Section -Post
 ExecWait '"$INSTDIR\${MAIN_APP_EXE}" service install'
 ExecWait '"$INSTDIR\${MAIN_APP_EXE}" service start'
@@ -326,9 +363,9 @@ DetailPrint "Deleting application files..."
 Delete "$INSTDIR\${UI_APP_EXE}"
 Delete "$INSTDIR\${MAIN_APP_EXE}"
 Delete "$INSTDIR\wintun.dll"
-!if ${ARCH} == "amd64"
+# Legacy: pre-Wails installs shipped opengl32.dll (Mesa3D for Fyne); remove
+# any leftover copy on uninstall so old upgrades don't leave it behind.
 Delete "$INSTDIR\opengl32.dll"
-!endif
 DetailPrint "Removing application directory..."
 RmDir /r "$INSTDIR"
 

client/internal/acl/manager.go

@@ -11,7 +11,6 @@ import (
 	"time"
 
 	"github.com/hashicorp/go-multierror"
-	"github.com/mitchellh/hashstructure/v2"
 	log "github.com/sirupsen/logrus"
 
 	nberrors "github.com/netbirdio/netbird/client/errors"
@@ -31,13 +30,11 @@ type Manager interface {
 
 // DefaultManager uses firewall manager to handle
 type DefaultManager struct {
-	firewall           firewall.Manager
-	ipsetCounter       int
-	peerRulesPairs     map[id.RuleID][]firewall.Rule
-	routeRules         map[id.RuleID]struct{}
-	previousConfigHash uint64
-	hasAppliedConfig   bool
-	mutex              sync.Mutex
+	firewall       firewall.Manager
+	ipsetCounter   int
+	peerRulesPairs map[id.RuleID][]firewall.Rule
+	routeRules     map[id.RuleID]struct{}
+	mutex          sync.Mutex
 }
 
 func NewDefaultManager(fm firewall.Manager) *DefaultManager {
@@ -60,23 +57,6 @@ func (d *DefaultManager) ApplyFiltering(networkMap *mgmProto.NetworkMap, dnsRout
 		return
 	}
 
-	// Skip the full rebuild + flush when the inputs that drive the firewall
-	// state are byte-for-byte identical to the last successfully applied
-	// update. Management re-sends the same network map far more often than it
-	// actually changes (account-wide updates, peer meta churn), and rebuilding
-	// every peer/route ACL and flushing the firewall on every such sync is the
-	// dominant client-side cost when nothing changed. Mirrors the same guard the
-	// DNS server already uses (previousConfigHash). Only the fields ApplyFiltering
-	// consumes participate in the hash, so an unrelated map change cannot mask a
-	// real ACL change.
-	hash, err := d.firewallConfigHash(networkMap, dnsRouteFeatureFlag)
-	if err != nil {
-		log.Errorf("unable to hash firewall configuration, applying unconditionally: %v", err)
-	} else if d.hasAppliedConfig && d.previousConfigHash == hash {
-		log.Debugf("not applying the firewall configuration update as there is nothing new (hash: %d)", hash)
-		return
-	}
-
 	start := time.Now()
 	defer func() {
 		total := 0
@@ -90,49 +70,13 @@ func (d *DefaultManager) ApplyFiltering(networkMap *mgmProto.NetworkMap, dnsRout
 
 	d.applyPeerACLs(networkMap)
 
-	routeErr := d.applyRouteACLs(networkMap.RoutesFirewallRules, dnsRouteFeatureFlag)
-	if routeErr != nil {
-		log.Errorf("Failed to apply route ACLs: %v", routeErr)
+	if err := d.applyRouteACLs(networkMap.RoutesFirewallRules, dnsRouteFeatureFlag); err != nil {
+		log.Errorf("Failed to apply route ACLs: %v", err)
 	}
 
-	flushErr := d.firewall.Flush()
-	if flushErr != nil {
-		log.Error("failed to flush firewall rules: ", flushErr)
+	if err := d.firewall.Flush(); err != nil {
+		log.Error("failed to flush firewall rules: ", err)
 	}
-
-	// Only remember the hash once the firewall actually reflects this config.
-	// If applying or flushing failed, leave the previous hash untouched so the
-	// next (possibly identical) update is not skipped and gets a chance to
-	// reconcile the firewall state.
-	if err == nil && routeErr == nil && flushErr == nil {
-		d.previousConfigHash = hash
-		d.hasAppliedConfig = true
-	} else {
-		d.hasAppliedConfig = false
-	}
-}
-
-// firewallConfigHash hashes exactly the inputs ApplyFiltering uses to build the
-// firewall state, so an identical hash means an identical resulting ruleset.
-func (d *DefaultManager) firewallConfigHash(networkMap *mgmProto.NetworkMap, dnsRouteFeatureFlag bool) (uint64, error) {
-	return hashstructure.Hash(struct {
-		PeerRules            []*mgmProto.FirewallRule
-		PeerRulesIsEmpty     bool
-		RouteRules           []*mgmProto.RouteFirewallRule
-		RouteRulesIsEmpty    bool
-		DNSRouteFeatureFlag  bool
-	}{
-		PeerRules:           networkMap.GetFirewallRules(),
-		PeerRulesIsEmpty:    networkMap.GetFirewallRulesIsEmpty(),
-		RouteRules:          networkMap.GetRoutesFirewallRules(),
-		RouteRulesIsEmpty:   networkMap.GetRoutesFirewallRulesIsEmpty(),
-		DNSRouteFeatureFlag: dnsRouteFeatureFlag,
-	}, hashstructure.FormatV2, &hashstructure.HashOptions{
-		ZeroNil:         true,
-		IgnoreZeroValue: true,
-		SlicesAsSets:    true,
-		UseStringer:     true,
-	})
 }
 
 func (d *DefaultManager) applyPeerACLs(networkMap *mgmProto.NetworkMap) {

client/internal/acl/manager_test.go

@@ -1,7 +1,6 @@
 package acl
 
 import (
-	"fmt"
 	"net/netip"
 	"testing"
 
@@ -486,149 +485,3 @@ func TestPortInfoEmpty(t *testing.T) {
 		})
 	}
 }
-
-// TestApplyFilteringSkipsUnchangedConfig verifies that an identical network map
-// re-applied is recognized as a no-op (hash unchanged), while a real change to
-// any firewall-relevant input forces a re-apply (hash changes). This is the
-// guard that prevents a full ruleset rebuild + flush on every redundant sync.
-func TestApplyFilteringSkipsUnchangedConfig(t *testing.T) {
-	t.Setenv("NB_WG_KERNEL_DISABLED", "true")
-	t.Setenv(firewall.EnvForceUserspaceFirewall, "true")
-
-	ctrl := gomock.NewController(t)
-	defer ctrl.Finish()
-
-	ifaceMock := mocks.NewMockIFaceMapper(ctrl)
-	ifaceMock.EXPECT().IsUserspaceBind().Return(true).AnyTimes()
-	ifaceMock.EXPECT().SetFilter(gomock.Any())
-	network := netip.MustParsePrefix("172.0.0.1/32")
-	ifaceMock.EXPECT().Name().Return("lo").AnyTimes()
-	ifaceMock.EXPECT().Address().Return(wgaddr.Address{
-		IP:      network.Addr(),
-		Network: network,
-	}).AnyTimes()
-	ifaceMock.EXPECT().GetWGDevice().Return(nil).AnyTimes()
-
-	fw, err := firewall.NewFirewall(ifaceMock, nil, flowLogger, false, iface.DefaultMTU)
-	require.NoError(t, err)
-	defer func() {
-		require.NoError(t, fw.Close(nil))
-	}()
-
-	acl := NewDefaultManager(fw)
-
-	networkMap := &mgmProto.NetworkMap{
-		FirewallRules: []*mgmProto.FirewallRule{
-			{
-				PeerIP:    "10.93.0.1",
-				Direction: mgmProto.RuleDirection_IN,
-				Action:    mgmProto.RuleAction_ACCEPT,
-				Protocol:  mgmProto.RuleProtocol_TCP,
-				Port:      "22",
-			},
-		},
-		FirewallRulesIsEmpty: false,
-	}
-
-	acl.ApplyFiltering(networkMap, false)
-	require.True(t, acl.hasAppliedConfig, "config should be marked applied after first apply")
-	firstHash := acl.previousConfigHash
-	require.NotZero(t, firstHash)
-
-	// Re-applying the identical map must not change the recorded hash: the
-	// expensive rebuild path was skipped.
-	acl.ApplyFiltering(networkMap, false)
-	assert.Equal(t, firstHash, acl.previousConfigHash,
-		"identical re-apply must be a no-op (hash unchanged)")
-
-	// A real change must produce a different hash and re-apply.
-	networkMap.FirewallRules[0].Action = mgmProto.RuleAction_DROP
-	acl.ApplyFiltering(networkMap, false)
-	assert.NotEqual(t, firstHash, acl.previousConfigHash,
-		"changing a rule's action must force a re-apply (hash changed)")
-
-	// The dnsRouteFeatureFlag also participates in the hash.
-	changedHash := acl.previousConfigHash
-	acl.ApplyFiltering(networkMap, true)
-	assert.NotEqual(t, changedHash, acl.previousConfigHash,
-		"flipping dnsRouteFeatureFlag must force a re-apply (hash changed)")
-}
-
-func buildNetworkMap(peerRules, routeRules int) *mgmProto.NetworkMap {
-	nm := &mgmProto.NetworkMap{
-		FirewallRulesIsEmpty:      peerRules == 0,
-		RoutesFirewallRulesIsEmpty: routeRules == 0,
-	}
-	for i := range peerRules {
-		nm.FirewallRules = append(nm.FirewallRules, &mgmProto.FirewallRule{
-			PeerIP:    fmt.Sprintf("10.%d.%d.%d", i>>16&0xff, i>>8&0xff, i&0xff),
-			Direction: mgmProto.RuleDirection_IN,
-			Action:    mgmProto.RuleAction_ACCEPT,
-			Protocol:  mgmProto.RuleProtocol_TCP,
-			Port:      fmt.Sprintf("%d", 1024+i%64511),
-		})
-	}
-	for i := range routeRules {
-		nm.RoutesFirewallRules = append(nm.RoutesFirewallRules, &mgmProto.RouteFirewallRule{
-			Destination:  fmt.Sprintf("192.168.%d.0/24", i%256),
-			SourceRanges: []string{fmt.Sprintf("10.0.%d.0/24", i%256)},
-			Action:       mgmProto.RuleAction_ACCEPT,
-			Protocol:     mgmProto.RuleProtocol_ALL,
-		})
-	}
-	return nm
-}
-
-func BenchmarkFirewallConfigHash_Small(b *testing.B) {
-	d := &DefaultManager{}
-	nm := buildNetworkMap(10, 5)
-	b.ResetTimer()
-	for b.Loop() {
-		_, _ = d.firewallConfigHash(nm, false)
-	}
-}
-
-func BenchmarkFirewallConfigHash_Medium(b *testing.B) {
-	d := &DefaultManager{}
-	nm := buildNetworkMap(100, 50)
-	b.ResetTimer()
-	for b.Loop() {
-		_, _ = d.firewallConfigHash(nm, false)
-	}
-}
-
-func BenchmarkFirewallConfigHash_Large(b *testing.B) {
-	d := &DefaultManager{}
-	nm := buildNetworkMap(1000, 200)
-	b.ResetTimer()
-	for b.Loop() {
-		_, _ = d.firewallConfigHash(nm, false)
-	}
-}
-
-// TestFirewallConfigHashDeterministic verifies the hash is stable for equal
-// inputs and order-independent for the rule slices (management does not
-// guarantee rule order).
-func TestFirewallConfigHashDeterministic(t *testing.T) {
-	d := &DefaultManager{}
-
-	nm1 := &mgmProto.NetworkMap{
-		FirewallRules: []*mgmProto.FirewallRule{
-			{PeerIP: "10.0.0.1", Direction: mgmProto.RuleDirection_IN, Action: mgmProto.RuleAction_ACCEPT, Protocol: mgmProto.RuleProtocol_TCP, Port: "22"},
-			{PeerIP: "10.0.0.2", Direction: mgmProto.RuleDirection_IN, Action: mgmProto.RuleAction_DROP, Protocol: mgmProto.RuleProtocol_TCP, Port: "80"},
-		},
-	}
-	// Same rules, reversed order.
-	nm2 := &mgmProto.NetworkMap{
-		FirewallRules: []*mgmProto.FirewallRule{
-			nm1.FirewallRules[1],
-			nm1.FirewallRules[0],
-		},
-	}
-
-	h1, err := d.firewallConfigHash(nm1, false)
-	require.NoError(t, err)
-	h2, err := d.firewallConfigHash(nm2, false)
-	require.NoError(t, err)
-	assert.Equal(t, h1, h2, "hash must be order-independent for rule slices")
-}

client/internal/auth/auth.go

@@ -3,6 +3,7 @@ package auth
 import (
 	"context"
 	"net/url"
+	"strings"
 	"sync"
 	"time"
 
@@ -21,6 +22,25 @@ import (
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
 )
 
+// peerLoginExpiredMsg is the exact phrase the management server returns
+// when a previously SSO-enrolled peer's login has expired. Sourced from
+// shared/management/status/error.go (NewPeerLoginExpiredError). Matched
+// by substring so a future server-side rewording that keeps the phrase
+// still triggers the friendly fallback in Login().
+const peerLoginExpiredMsg = "peer login has expired"
+
+// errSetupKeyOnSSOExpiredPeer replaces the raw management error when the
+// user runs `netbird login -k <setup-key>` against a peer that was
+// originally enrolled via SSO. Wrapped in a PermissionDenied gRPC status
+// so callers' existing isPermissionDenied / isAuthError checks still
+// classify it correctly (early-exit from retry backoff, StatusNeedsLogin
+// in the server state machine).
+var errSetupKeyOnSSOExpiredPeer = status.Error(
+	codes.PermissionDenied,
+	"this peer was originally enrolled via SSO and its session has expired. "+
+		"Setup keys can only enrol new peers — run `netbird up` (interactive SSO) to re-login.",
+)
+
 // Auth manages authentication operations with the management server
 // It maintains a long-lived connection and automatically handles reconnection with backoff
 type Auth struct {
@@ -184,6 +204,15 @@ func (a *Auth) Login(ctx context.Context, setupKey string, jwtToken string) (err
 			log.Debugf("peer registration required")
 			_, err = a.registerPeer(client, ctx, setupKey, jwtToken, pubSSHKey)
 			if err != nil {
+				// The peer pub-key is already on file with the management
+				// server (originally enrolled via SSO) and the session has
+				// expired. The setup-key path can only enrol new peers, so
+				// retrying with -k will keep failing. Replace the raw mgm
+				// message with an actionable hint that tells the user to
+				// re-authenticate via SSO instead.
+				if setupKey != "" && jwtToken == "" && isPeerLoginExpired(err) {
+					err = errSetupKeyOnSSOExpiredPeer
+				}
 				isAuthError = isPermissionDenied(err)
 				return err
 			}
@@ -474,3 +503,16 @@ func isLoginNeeded(err error) bool {
 func isRegistrationNeeded(err error) bool {
 	return isPermissionDenied(err)
 }
+
+// isPeerLoginExpired reports whether err is the management server's
+// "peer login has expired" PermissionDenied response. Used by Login to
+// detect the case where the caller passed a setup-key but the peer is
+// actually an SSO-enrolled record whose session needs refreshing — the
+// setup-key path cannot help there.
+func isPeerLoginExpired(err error) bool {
+	if !isPermissionDenied(err) {
+		return false
+	}
+	s, _ := status.FromError(err)
+	return strings.Contains(s.Message(), peerLoginExpiredMsg)
+}

client/internal/auth/auth_test.go

@@ -0,0 +1,80 @@
+package auth
+
+import (
+	"errors"
+	"strings"
+	"testing"
+
+	"google.golang.org/grpc/codes"
+	"google.golang.org/grpc/status"
+)
+
+func TestIsPeerLoginExpired(t *testing.T) {
+	cases := []struct {
+		name string
+		err  error
+		want bool
+	}{
+		{
+			name: "nil",
+			err:  nil,
+			want: false,
+		},
+		{
+			name: "plain error (not a gRPC status)",
+			err:  errors.New("network read: connection reset"),
+			want: false,
+		},
+		{
+			name: "PermissionDenied with different message",
+			err:  status.Error(codes.PermissionDenied, "user is blocked"),
+			want: false,
+		},
+		{
+			name: "Unauthenticated with the expected phrase",
+			// Wrong status code — must still return false.
+			err:  status.Error(codes.Unauthenticated, "peer login has expired, please log in once more"),
+			want: false,
+		},
+		{
+			name: "exact server message",
+			err:  status.Error(codes.PermissionDenied, "peer login has expired, please log in once more"),
+			want: true,
+		},
+		{
+			name: "phrase as substring",
+			// Future-proofing: if mgm reworords but keeps the phrase,
+			// the friendly fallback must still kick in.
+			err:  status.Error(codes.PermissionDenied, "session refused: peer login has expired (account=foo)"),
+			want: true,
+		},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := isPeerLoginExpired(tc.err); got != tc.want {
+				t.Fatalf("isPeerLoginExpired(%v) = %v, want %v", tc.err, got, tc.want)
+			}
+		})
+	}
+}
+
+func TestErrSetupKeyOnSSOExpiredPeer(t *testing.T) {
+	// Sentinel must surface as PermissionDenied so the upstream
+	// isPermissionDenied / isAuthError checks classify it correctly
+	// (short-circuit retry backoff, set StatusNeedsLogin).
+	if !isPermissionDenied(errSetupKeyOnSSOExpiredPeer) {
+		t.Fatalf("errSetupKeyOnSSOExpiredPeer must be a PermissionDenied gRPC error")
+	}
+
+	// Message must actually mention SSO and `netbird up` so it is
+	// actionable for the end user. Loose substring checks keep the
+	// test resilient to copy edits.
+	s, _ := status.FromError(errSetupKeyOnSSOExpiredPeer)
+	msg := strings.ToLower(s.Message())
+	for _, want := range []string{"sso", "netbird up"} {
+		if !strings.Contains(msg, want) {
+			t.Errorf("sentinel message should contain %q, got %q", want, s.Message())
+		}
+	}
+}

client/internal/auth/pending_flow.go

@@ -0,0 +1,89 @@
+package auth
+
+import (
+	"context"
+	"sync"
+	"time"
+)
+
+// PendingFlow stores an in-progress OAuth flow between the RPC that
+// initiates it (returns the verification URI to the UI) and the RPC
+// that waits for the user to complete it. The flow handle, the
+// device-code info, and the absolute expiry are kept together so the
+// waiting RPC can validate the device code and reuse the same flow.
+//
+// PendingFlow is safe for concurrent use; callers must not access the
+// stored fields directly.
+type PendingFlow struct {
+	mu         sync.Mutex
+	flow       OAuthFlow
+	info       AuthFlowInfo
+	expiresAt  time.Time
+	waitCancel context.CancelFunc
+}
+
+// NewPendingFlow returns an empty PendingFlow ready to be populated by Set.
+func NewPendingFlow() *PendingFlow {
+	return &PendingFlow{}
+}
+
+// Set stores the flow and its authorization info, computing the absolute
+// expiry from info.ExpiresIn (seconds, as returned by the IdP).
+func (p *PendingFlow) Set(flow OAuthFlow, info AuthFlowInfo) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.flow = flow
+	p.info = info
+	p.expiresAt = time.Now().Add(time.Duration(info.ExpiresIn) * time.Second)
+}
+
+// Get returns the stored flow, info, and whether a flow is currently
+// pending. Returns (nil, zero, false) after Clear or before Set.
+func (p *PendingFlow) Get() (OAuthFlow, AuthFlowInfo, bool) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	if p.flow == nil {
+		return nil, AuthFlowInfo{}, false
+	}
+	return p.flow, p.info, true
+}
+
+// ExpiresAt returns the absolute expiry of the pending flow. Returns
+// the zero time when no flow is pending.
+func (p *PendingFlow) ExpiresAt() time.Time {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	return p.expiresAt
+}
+
+// SetWaitCancel records the cancel function for the goroutine currently
+// blocked in WaitToken so a new RequestAuth can preempt it.
+func (p *PendingFlow) SetWaitCancel(cancel context.CancelFunc) {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.waitCancel = cancel
+}
+
+// CancelWait invokes and clears the stored wait-cancel, if any. Safe to
+// call when no wait is in progress.
+func (p *PendingFlow) CancelWait() {
+	p.mu.Lock()
+	cancel := p.waitCancel
+	p.waitCancel = nil
+	p.mu.Unlock()
+	if cancel != nil {
+		cancel()
+	}
+}
+
+// Clear resets the pending flow to empty. Any stored wait-cancel is
+// dropped without being invoked — call CancelWait first if the waiting
+// goroutine must be stopped.
+func (p *PendingFlow) Clear() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	p.flow = nil
+	p.info = AuthFlowInfo{}
+	p.expiresAt = time.Time{}
+	p.waitCancel = nil
+}

client/internal/auth/sessionwatch/event.go

@@ -0,0 +1,82 @@
+package sessionwatch
+
+import (
+	"strconv"
+	"time"
+)
+
+// internal event kinds are no longer exposed: the watcher drives the Sink
+// directly (NotifyStateChange on deadline change/clear, PublishEvent at
+// each warning lead). Tests use a mock Sink to observe what the watcher
+// emits.
+
+// Metadata keys attached by the daemon to session-warning SystemEvents.
+// The UI tray reads these to build a locale-aware notification without
+// relying on the daemon's locale-less UserMessage string, and to
+// disambiguate the T-WarningLead notification from the T-FinalWarningLead
+// fallback that auto-opens the SessionAboutToExpire dialog.
+const (
+	// MetaSessionWarning is set to "true" on both warning events (T-10 and
+	// T-2) so the UI can detect a session-warning SystemEvent without
+	// matching on the message text. Use MetaSessionFinal to distinguish
+	// the two.
+	MetaSessionWarning = "session_warning"
+	// MetaSessionFinal is set to "true" on the T-FinalWarningLead event
+	// only. Consumers that need to auto-open the SessionAboutToExpire
+	// dialog gate on this; T-WarningLead events leave the field unset.
+	MetaSessionFinal = "session_final_warning"
+	// MetaSessionExpiresAt carries the absolute UTC deadline encoded with
+	// FormatExpiresAt; consumers must decode with ParseExpiresAt so a
+	// future format change stays a single edit.
+	MetaSessionExpiresAt = "session_expires_at"
+	// MetaSessionLeadMinutes carries the lead in whole minutes (WarningLead
+	// for the T-10 event, FinalWarningLead for the T-2 event) so the UI
+	// can show "expires in ~N minutes" without hardcoding either constant.
+	MetaSessionLeadMinutes = "lead_minutes"
+	// MetaSessionDeadlineRejected is attached to the ERROR/AUTHENTICATION
+	// SystemEvent the daemon emits when it discards a deadline from the
+	// management server (pre-epoch, too far in the future, or past the
+	// clock-skew tolerance). The value is the rejection reason string.
+	// userMessage is left empty; the UI detects the event via this key
+	// and builds a localized notification — same pattern as the session
+	// warnings above.
+	MetaSessionDeadlineRejected = "session_deadline_rejected"
+)
+
+// expiresAtLayout is the wire format used for MetaSessionExpiresAt.
+// Producer and consumers both go through FormatExpiresAt/ParseExpiresAt
+// so this layout stays a single source of truth.
+const expiresAtLayout = time.RFC3339
+
+// FormatExpiresAt encodes a deadline for MetaSessionExpiresAt. Always
+// emits UTC so a consumer in another timezone reads the same wall-clock
+// deadline.
+func FormatExpiresAt(t time.Time) string {
+	return t.UTC().Format(expiresAtLayout)
+}
+
+// ParseExpiresAt decodes the MetaSessionExpiresAt value back to a UTC
+// time. Returns an error when the field is empty or malformed; the
+// caller decides whether to fall back (zero value) or propagate.
+func ParseExpiresAt(s string) (time.Time, error) {
+	t, err := time.Parse(expiresAtLayout, s)
+	if err != nil {
+		return time.Time{}, err
+	}
+	return t.UTC(), nil
+}
+
+// FormatLeadMinutes encodes a lead duration for MetaSessionLeadMinutes
+// as the integer count of whole minutes. Sub-minute residuals are
+// truncated — the field is informational ("expires in ~N minutes") and
+// fractional minutes don't change what the UI displays.
+func FormatLeadMinutes(d time.Duration) string {
+	return strconv.Itoa(int(d / time.Minute))
+}
+
+// ParseLeadMinutes decodes a MetaSessionLeadMinutes value. Returns 0
+// and the parse error for malformed input; consumers that prefer a
+// silent fallback can simply ignore the error.
+func ParseLeadMinutes(s string) (int, error) {
+	return strconv.Atoi(s)
+}

client/internal/auth/sessionwatch/watcher.go

@@ -0,0 +1,387 @@
+// Package sessionwatch tracks the SSO session expiry deadline that the
+// management server publishes via LoginResponse / SyncResponse and fires
+// two warning events at fixed lead times before expiry: an interactive
+// T-WarningLead notification and a dismiss-gated T-FinalWarningLead
+// fallback dialog.
+//
+// The watcher is idempotent: Update may be called as often as the network
+// map snapshots arrive. Repeating the same deadline is a no-op; a new
+// deadline reschedules the timers and arms a fresh warning cycle.
+//
+// Warning firing is edge-detected. Each unique deadline value fires each
+// warning callback at most once.
+package sessionwatch
+
+import (
+	"errors"
+	"fmt"
+	"sync"
+	"time"
+
+	log "github.com/sirupsen/logrus"
+
+	cProto "github.com/netbirdio/netbird/client/proto"
+)
+
+const (
+	// Skew tolerates a small clock difference between the management
+	// server and this peer before treating a deadline as "in the past".
+	// Slightly above typical NTP drift; tight enough that the UI doesn't
+	// paint a stale expiry as if it were valid.
+	Skew = 30 * time.Second
+
+	// maxDeadlineHorizon caps how far in the future an accepted deadline
+	// can sit. A timestamp beyond this is almost certainly a protocol
+	// glitch, and silently arming a 100-year timer would hide the bug.
+	maxDeadlineHorizon = 10 * 365 * 24 * time.Hour
+
+	// WarningLead is how far before expiry the first (interactive)
+	// warning fires. Drives the T-10 OS notification with
+	// Extend/Dismiss actions.
+	WarningLead = 10 * time.Minute
+
+	// FinalWarningLead is how far before expiry the fallback final
+	// warning fires. Drives the auto-opened SessionAboutToExpire dialog,
+	// but only when the user has not dismissed the T-WarningLead warning
+	// for the same deadline. Must be strictly less than WarningLead.
+	FinalWarningLead = 2 * time.Minute
+)
+
+var (
+	// ErrDeadlineBeforeEpoch is returned by Update when the supplied
+	// deadline pre-dates 1970-01-01.
+	ErrDeadlineBeforeEpoch = errors.New("session deadline before unix epoch")
+
+	// ErrDeadlineTooFarFuture is returned by Update when the supplied
+	// deadline is more than maxDeadlineHorizon in the future.
+	ErrDeadlineTooFarFuture = errors.New("session deadline too far in the future")
+
+	// ErrDeadlineInPast is returned by Update when the supplied deadline
+	// is more than Skew in the past.
+	ErrDeadlineInPast = errors.New("session deadline in the past")
+)
+
+// StatusRecorder is the side-effect surface the watcher drives on every
+// state transition. Production wires this to peer.Status (SetSessionExpiresAt
+// for deadline change/clear, PublishEvent for the two warnings); tests pass
+// a fake recorder so the same surface is observable without an engine.
+//
+// The watcher is the single owner of the deadline propagated to the
+// recorder: every set, clear, sanity-check rejection and Close routes the
+// value through SetSessionExpiresAt, so the SubscribeStatus snapshot the UI
+// reads can never drift from the watcher's timer state. (SetSessionExpiresAt
+// fans out its own state-change notification, so no separate notify is
+// needed.) The recorder is server-scoped and outlives this engine-scoped
+// watcher — without the Close-time clear a teardown (Down, or the Down+Up of
+// a profile switch) would leave the next session showing the previous one's
+// stale "expires in" value.
+//
+// PublishEvent's signature mirrors peer.Status.PublishEvent: the watcher
+// composes the metadata internally so the wire format (MetaSession*) is
+// owned by sessionwatch, not the caller.
+type StatusRecorder interface {
+	SetSessionExpiresAt(deadline time.Time)
+	PublishEvent(
+		severity cProto.SystemEvent_Severity,
+		category cProto.SystemEvent_Category,
+		message string,
+		userMessage string,
+		metadata map[string]string,
+	)
+}
+
+// Watcher observes the latest session deadline and fires two warnings
+// before it expires: the interactive T-WarningLead notification, and the
+// fallback T-FinalWarningLead dialog (suppressed when the user dismissed
+// the first one for the same deadline). Safe for concurrent use.
+type Watcher struct {
+	lead      time.Duration
+	finalLead time.Duration
+
+	mu           sync.Mutex
+	current      time.Time
+	timer        *time.Timer
+	finalTimer   *time.Timer
+	firedAt      time.Time // deadline value the T-WarningLead callback last fired against
+	finalFiredAt time.Time // deadline value the T-FinalWarningLead callback last fired against
+	dismissedAt  time.Time // deadline value the user dismissed via Dismiss(); gates fireFinal
+	closed       bool
+	recorder     StatusRecorder
+}
+
+// New returns a watcher with the package defaults WarningLead and
+// FinalWarningLead. Pass nil for recorder to silence side effects (handy
+// in unit tests that exercise sanity checks without observing the publish
+// path).
+func New(recorder StatusRecorder) *Watcher {
+	return NewWithLeads(WarningLead, FinalWarningLead, recorder)
+}
+
+// NewWithLeads returns a watcher with custom lead times. Useful for tests.
+// final must be strictly less than lead; otherwise both timers fire in the
+// wrong order or simultaneously and the UI flow breaks. A zero final lead
+// disables the final-warning timer entirely (see armTimerLocked) so a
+// millisecond-scale deadline doesn't flush both timers in one tick.
+func NewWithLeads(lead, final time.Duration, recorder StatusRecorder) *Watcher {
+	return &Watcher{
+		lead:      lead,
+		finalLead: final,
+		recorder:  recorder,
+	}
+}
+
+// Update sets the latest deadline. Pass the zero time to clear (e.g. when
+// a Sync push from the server omits the field because login expiration
+// was disabled).
+//
+// Same-value updates are no-ops. A different non-zero value cancels any
+// pending timer, resets the "already fired" guard, and arms a new one.
+//
+// Returns one of the sentinel Err* values when the deadline fails the
+// sanity checks (pre-epoch, far future, or in the past beyond Skew).
+// In every error case the watcher first clears its state so it stays
+// consistent with what the caller will push into its other sinks (e.g.
+// applySessionDeadline forces a zero deadline into the status recorder
+// after a non-nil error).
+func (w *Watcher) Update(deadline time.Time) error {
+	w.mu.Lock()
+	if w.closed {
+		w.mu.Unlock()
+		return nil
+	}
+
+	if deadline.IsZero() {
+		w.clearLocked()
+		return nil
+	}
+
+	now := time.Now()
+	switch {
+	case deadline.Before(time.Unix(0, 0)):
+		w.clearLocked()
+		return fmt.Errorf("%w: %v", ErrDeadlineBeforeEpoch, deadline)
+	case deadline.After(now.Add(maxDeadlineHorizon)):
+		w.clearLocked()
+		return fmt.Errorf("%w: %v", ErrDeadlineTooFarFuture, deadline)
+	case deadline.Before(now.Add(-Skew)):
+		w.clearLocked()
+		return fmt.Errorf("%w: %v (now=%v)", ErrDeadlineInPast, deadline, now)
+	}
+
+	if deadline.Equal(w.current) {
+		w.mu.Unlock()
+		return nil
+	}
+
+	w.stopTimerLocked()
+	w.current = deadline
+	// Reset every per-deadline guard so a refreshed deadline arms a fresh
+	// warning cycle: both edge triggers and the user Dismiss decision
+	// (the user agreed to the old deadline expiring; a new deadline
+	// restarts the contract).
+	w.firedAt = time.Time{}
+	w.finalFiredAt = time.Time{}
+	w.dismissedAt = time.Time{}
+
+	w.armTimerLocked(deadline)
+	recorder := w.recorder
+	w.mu.Unlock()
+	if recorder != nil {
+		recorder.SetSessionExpiresAt(deadline)
+	}
+	log.Infof("auth session deadline set to: %s (in %s)", deadline.Format(time.RFC3339), time.Until(deadline).Round(time.Second))
+	return nil
+}
+
+// Deadline returns the most recently observed deadline. Zero when no
+// deadline is currently tracked.
+func (w *Watcher) Deadline() time.Time {
+	w.mu.Lock()
+	defer w.mu.Unlock()
+	return w.current
+}
+
+// Dismiss records the user's "Dismiss" action against the current deadline
+// and suppresses the upcoming final-warning callback for that deadline.
+// Idempotent: repeated calls are no-ops. A subsequent Update with a fresh
+// deadline resets the dismissal so the final-warning cycle re-arms.
+//
+// No-op when the watcher holds no deadline or has been closed.
+func (w *Watcher) Dismiss() {
+	w.mu.Lock()
+	defer w.mu.Unlock()
+	if w.closed || w.current.IsZero() {
+		return
+	}
+	if w.dismissedAt.Equal(w.current) {
+		return
+	}
+	w.dismissedAt = w.current
+	// Cancel the armed final-warning timer eagerly. fireFinal would also
+	// gate on dismissedAt, but stopping the timer avoids a wakeup with
+	// nothing to do and makes the intent visible.
+	if w.finalTimer != nil {
+		w.finalTimer.Stop()
+		w.finalTimer = nil
+	}
+	log.Infof("auth session final-warning dismissed for deadline %s", w.current.Format(time.RFC3339))
+}
+
+// Close stops any pending timer and drops the deadline on the status
+// recorder. Update calls after Close are ignored. Clearing the recorder
+// here is what keeps a teardown (Down, or the Down+Up of a profile switch)
+// from leaving the next session showing this one's stale "expires in"
+// value — the recorder is server-scoped and outlives this engine-scoped
+// watcher, so nothing else drops the anchor on teardown.
+func (w *Watcher) Close() {
+	w.mu.Lock()
+	if w.closed {
+		w.mu.Unlock()
+		return
+	}
+	w.closed = true
+	w.stopTimerLocked()
+	hadDeadline := !w.current.IsZero()
+	w.current = time.Time{}
+	w.firedAt = time.Time{}
+	w.finalFiredAt = time.Time{}
+	w.dismissedAt = time.Time{}
+	recorder := w.recorder
+	w.mu.Unlock()
+	if recorder != nil && hadDeadline {
+		recorder.SetSessionExpiresAt(time.Time{})
+	}
+}
+
+// clearLocked drops the tracked deadline and notifies the recorder so
+// downstream consumers (SubscribeStatus stream, UI) drop their anchor.
+// The caller must hold w.mu; this helper releases it before invoking
+// the recorder.
+func (w *Watcher) clearLocked() {
+	if w.current.IsZero() {
+		w.mu.Unlock()
+		return
+	}
+	w.stopTimerLocked()
+	w.current = time.Time{}
+	w.firedAt = time.Time{}
+	w.finalFiredAt = time.Time{}
+	w.dismissedAt = time.Time{}
+	recorder := w.recorder
+	w.mu.Unlock()
+	if recorder != nil {
+		recorder.SetSessionExpiresAt(time.Time{})
+	}
+	log.Infof("auth session deadline cleared")
+}
+
+func (w *Watcher) stopTimerLocked() {
+	if w.timer != nil {
+		w.timer.Stop()
+		w.timer = nil
+	}
+	if w.finalTimer != nil {
+		w.finalTimer.Stop()
+		w.finalTimer = nil
+	}
+}
+
+func (w *Watcher) armTimerLocked(deadline time.Time) {
+	w.timer = armOneShotLocked(deadline.Add(-w.lead), func() { w.fire(deadline) })
+	// finalLead <= 0 disables the final-warning timer entirely. Used by
+	// tests that predate the final-warning fallback so a millisecond-scale
+	// deadline does not flush both timers at once.
+	if w.finalLead > 0 {
+		w.finalTimer = armOneShotLocked(deadline.Add(-w.finalLead), func() { w.fireFinal(deadline) })
+	}
+}
+
+func (w *Watcher) fire(armedFor time.Time) {
+	w.mu.Lock()
+	if w.closed || !w.current.Equal(armedFor) {
+		// Deadline moved while we were waiting (e.g. a successful extend).
+		// The reschedule path armed a fresh timer; this one is stale.
+		w.mu.Unlock()
+		return
+	}
+	if !w.firedAt.IsZero() && w.firedAt.Equal(armedFor) {
+		w.mu.Unlock()
+		return
+	}
+	w.firedAt = armedFor
+	recorder := w.recorder
+	w.mu.Unlock()
+	if recorder == nil {
+		return
+	}
+	log.Infof("auth session expiry soon warning fired")
+	publishWarning(recorder, armedFor, false)
+}
+
+// fireFinal mirrors fire for the T-FinalWarningLead timer with an extra
+// dismiss-gate: if the user dismissed the T-WarningLead notification for
+// this deadline, the final warning is suppressed entirely.
+func (w *Watcher) fireFinal(armedFor time.Time) {
+	w.mu.Lock()
+	if w.closed || !w.current.Equal(armedFor) {
+		w.mu.Unlock()
+		return
+	}
+	if !w.finalFiredAt.IsZero() && w.finalFiredAt.Equal(armedFor) {
+		w.mu.Unlock()
+		return
+	}
+	if w.dismissedAt.Equal(armedFor) {
+		w.mu.Unlock()
+		log.Infof("auth session final-warning skipped (dismissed by user)")
+		return
+	}
+	w.finalFiredAt = armedFor
+	recorder := w.recorder
+	w.mu.Unlock()
+	if recorder == nil {
+		return
+	}
+	log.Infof("auth session final-warning fired")
+	publishWarning(recorder, armedFor, true)
+}
+
+// armOneShotLocked schedules cb at fireAt. When fireAt is already in the
+// past it dispatches on the next scheduler tick so a state-change recorder
+// notification (invoked after w.mu is released) lands first. Caller must
+// hold w.mu.
+func armOneShotLocked(fireAt time.Time, cb func()) *time.Timer {
+	delay := time.Until(fireAt)
+	if delay <= 0 {
+		return time.AfterFunc(0, cb)
+	}
+	return time.AfterFunc(delay, cb)
+}
+
+// publishWarning composes the SystemEvent for a watcher-fired warning and
+// pushes it through the recorder. Severity is CRITICAL on both — bypassing
+// the user's Notifications toggle is deliberate: missing the warning
+// window forces the post-mortem SessionExpired flow (tunnel torn down,
+// lock icon, manual re-login), which is the UX we are trying to avoid.
+func publishWarning(recorder StatusRecorder, deadline time.Time, final bool) {
+	lead := WarningLead
+	message := "session expiry warning"
+	meta := map[string]string{
+		MetaSessionWarning:   "true",
+		MetaSessionExpiresAt: FormatExpiresAt(deadline),
+	}
+	if final {
+		lead = FinalWarningLead
+		message = "session expiry final warning"
+		meta[MetaSessionFinal] = "true"
+	}
+	meta[MetaSessionLeadMinutes] = FormatLeadMinutes(lead)
+
+	recorder.PublishEvent(
+		cProto.SystemEvent_CRITICAL,
+		cProto.SystemEvent_AUTHENTICATION,
+		message,
+		"",
+		meta,
+	)
+}

client/internal/auth/sessionwatch/watcher_test.go

@@ -0,0 +1,519 @@
+package sessionwatch
+
+import (
+	"errors"
+	"sync"
+	"testing"
+	"time"
+
+	cProto "github.com/netbirdio/netbird/client/proto"
+)
+
+// fakeRecorder satisfies StatusRecorder and records every call so tests
+// can observe what the watcher emits. SetSessionExpiresAt and PublishEvent
+// land in the same ordered events slice (with the Kind distinguishing
+// them) so tests that care about ordering still work. lastDeadline holds
+// the most recent value passed to SetSessionExpiresAt so tests can assert
+// the recorder ended up cleared/set as expected.
+type fakeRecorder struct {
+	mu           sync.Mutex
+	events       []event
+	lastDeadline time.Time
+}
+
+type eventKind int
+
+const (
+	stateChange eventKind = iota
+	publish
+)
+
+type event struct {
+	kind eventKind
+	// Set only for publish events.
+	severity cProto.SystemEvent_Severity
+	category cProto.SystemEvent_Category
+	message  string
+	meta     map[string]string
+}
+
+// SetSessionExpiresAt mirrors peer.Status: a same-value write is a no-op,
+// a real change records the new value and fans out a state-change (the
+// production recorder calls notifyStateChange internally). The baseline
+// is the zero time, so an initial clear before any deadline is set emits
+// nothing — matching the real recorder.
+func (r *fakeRecorder) SetSessionExpiresAt(deadline time.Time) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	if r.lastDeadline.Equal(deadline) {
+		return
+	}
+	r.lastDeadline = deadline
+	r.events = append(r.events, event{kind: stateChange})
+}
+
+func (r *fakeRecorder) deadline() time.Time {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	return r.lastDeadline
+}
+
+func (r *fakeRecorder) PublishEvent(
+	severity cProto.SystemEvent_Severity,
+	category cProto.SystemEvent_Category,
+	message string,
+	_ string,
+	metadata map[string]string,
+) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	r.events = append(r.events, event{
+		kind:     publish,
+		severity: severity,
+		category: category,
+		message:  message,
+		meta:     metadata,
+	})
+}
+
+func (r *fakeRecorder) snapshot() []event {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	out := make([]event, len(r.events))
+	copy(out, r.events)
+	return out
+}
+
+func (e event) isFinalWarning() bool {
+	return e.kind == publish && e.meta[MetaSessionFinal] == "true"
+}
+
+func (e event) isWarning() bool {
+	return e.kind == publish && e.meta[MetaSessionWarning] == "true" && e.meta[MetaSessionFinal] != "true"
+}
+
+func countWhere(events []event, pred func(event) bool) int {
+	n := 0
+	for _, e := range events {
+		if pred(e) {
+			n++
+		}
+	}
+	return n
+}
+
+func waitForEvents(t *testing.T, r *fakeRecorder, want int) []event {
+	t.Helper()
+	deadline := time.Now().Add(500 * time.Millisecond)
+	for time.Now().Before(deadline) {
+		if got := r.snapshot(); len(got) >= want {
+			return got
+		}
+		time.Sleep(5 * time.Millisecond)
+	}
+	got := r.snapshot()
+	t.Fatalf("timed out waiting for %d events, got %d: %+v", want, len(got), got)
+	return nil
+}
+
+// newWatcher builds a watcher with the final timer disabled (finalLead=0),
+// matching the lead-only behaviour the pre-final-warning tests assume.
+func newWatcher(lead time.Duration, r *fakeRecorder) *Watcher {
+	return NewWithLeads(lead, 0, r)
+}
+
+func TestUpdateZeroBeforeAnythingIsNoop(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	_ = w.Update(time.Time{})
+
+	if got := r.snapshot(); len(got) != 0 {
+		t.Fatalf("expected no events on initial zero, got %+v", got)
+	}
+}
+
+func TestUpdateNonZeroFiresStateChange(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	d := time.Now().Add(time.Hour)
+	_ = w.Update(d)
+
+	events := waitForEvents(t, r, 1)
+	if events[0].kind != stateChange {
+		t.Fatalf("expected stateChange, got %+v", events[0])
+	}
+	if !w.Deadline().Equal(d) {
+		t.Fatalf("deadline mismatch: %v vs %v", w.Deadline(), d)
+	}
+}
+
+func TestSameDeadlineIsNoop(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	d := time.Now().Add(time.Hour)
+	_ = w.Update(d)
+	_ = w.Update(d)
+	_ = w.Update(d)
+
+	events := waitForEvents(t, r, 1)
+	if len(events) != 1 {
+		t.Fatalf("expected exactly 1 event for repeated same deadline, got %d: %+v", len(events), events)
+	}
+}
+
+func TestWarningFiresOnceWithinLeadWindow(t *testing.T) {
+	r := &fakeRecorder{}
+	lead := 50 * time.Millisecond
+	w := newWatcher(lead, r)
+	defer w.Close()
+
+	// Deadline 80ms out — warning should fire after ~30ms.
+	d := time.Now().Add(80 * time.Millisecond)
+	_ = w.Update(d)
+
+	events := waitForEvents(t, r, 2)
+	if events[0].kind != stateChange {
+		t.Fatalf("event[0] should be stateChange, got %+v", events[0])
+	}
+	if !events[1].isWarning() {
+		t.Fatalf("event[1] should be a warning publish, got %+v", events[1])
+	}
+}
+
+func TestWarningFiresImmediatelyWhenAlreadyInsideWindow(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(time.Hour, r) // lead > delta => fire immediately
+	defer w.Close()
+
+	d := time.Now().Add(10 * time.Millisecond)
+	_ = w.Update(d)
+
+	events := waitForEvents(t, r, 2)
+	if !events[1].isWarning() {
+		t.Fatalf("expected immediate warning publish, got %+v", events[1])
+	}
+}
+
+func TestNewDeadlineCancelsPriorTimer(t *testing.T) {
+	r := &fakeRecorder{}
+	lead := 50 * time.Millisecond
+	w := newWatcher(lead, r)
+	defer w.Close()
+
+	first := time.Now().Add(80 * time.Millisecond) // would fire warning ~30ms in
+	_ = w.Update(first)
+
+	// Replace with a far-future deadline before the warning fires.
+	time.Sleep(5 * time.Millisecond)
+	second := time.Now().Add(time.Hour)
+	_ = w.Update(second)
+
+	// Wait past when first's warning would have fired.
+	time.Sleep(80 * time.Millisecond)
+
+	if n := countWhere(r.snapshot(), event.isWarning); n != 0 {
+		t.Fatalf("warning fired for cancelled deadline: %+v", r.snapshot())
+	}
+}
+
+func TestRefreshAfterFireArmsNewWarning(t *testing.T) {
+	r := &fakeRecorder{}
+	lead := 30 * time.Millisecond
+	w := newWatcher(lead, r)
+	defer w.Close()
+
+	first := time.Now().Add(50 * time.Millisecond)
+	_ = w.Update(first)
+
+	// Wait for stateChange + warning of the first cycle.
+	waitForEvents(t, r, 2)
+
+	// Simulate a successful extend: brand new deadline.
+	second := time.Now().Add(60 * time.Millisecond)
+	_ = w.Update(second)
+
+	// 4 events total: stateChange, warning (first), stateChange, warning (second).
+	events := waitForEvents(t, r, 4)
+	if events[2].kind != stateChange {
+		t.Fatalf("event[2] should be stateChange for the new deadline, got %+v", events[2])
+	}
+	if !events[3].isWarning() {
+		t.Fatalf("event[3] should be a warning publish for the new deadline, got %+v", events[3])
+	}
+}
+
+func TestUpdateZeroAfterNonZeroClearsState(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(time.Hour, r)
+	defer w.Close()
+
+	d := time.Now().Add(2 * time.Hour)
+	_ = w.Update(d)
+	waitForEvents(t, r, 1)
+
+	_ = w.Update(time.Time{})
+
+	events := waitForEvents(t, r, 2)
+	if events[1].kind != stateChange {
+		t.Fatalf("expected stateChange on clear, got %+v", events[1])
+	}
+	if !w.Deadline().IsZero() {
+		t.Fatalf("Deadline should be zero after clear")
+	}
+}
+
+func TestUpdateRejectsBeforeEpoch(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	good := time.Now().Add(time.Hour)
+	if err := w.Update(good); err != nil {
+		t.Fatalf("seed Update: %v", err)
+	}
+
+	err := w.Update(time.Unix(-100, 0))
+	if !errors.Is(err, ErrDeadlineBeforeEpoch) {
+		t.Fatalf("want ErrDeadlineBeforeEpoch, got %v", err)
+	}
+	if !w.Deadline().IsZero() {
+		t.Fatalf("rejected pre-epoch update must clear deadline; got %v", w.Deadline())
+	}
+}
+
+func TestUpdateRejectsTooFarFuture(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	good := time.Now().Add(time.Hour)
+	if err := w.Update(good); err != nil {
+		t.Fatalf("seed Update: %v", err)
+	}
+
+	err := w.Update(time.Now().Add(50 * 365 * 24 * time.Hour))
+	if !errors.Is(err, ErrDeadlineTooFarFuture) {
+		t.Fatalf("want ErrDeadlineTooFarFuture, got %v", err)
+	}
+	if !w.Deadline().IsZero() {
+		t.Fatalf("rejected far-future update must clear deadline; got %v", w.Deadline())
+	}
+}
+
+func TestUpdateInPastClearsDeadline(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	good := time.Now().Add(time.Hour)
+	if err := w.Update(good); err != nil {
+		t.Fatalf("seed Update: %v", err)
+	}
+	// Drain the stateChange from the seed.
+	waitForEvents(t, r, 1)
+
+	err := w.Update(time.Now().Add(-1 * time.Hour))
+	if !errors.Is(err, ErrDeadlineInPast) {
+		t.Fatalf("want ErrDeadlineInPast, got %v", err)
+	}
+	if !w.Deadline().IsZero() {
+		t.Fatalf("in-past update must clear the deadline, got %v", w.Deadline())
+	}
+	events := waitForEvents(t, r, 2)
+	if events[1].kind != stateChange {
+		t.Fatalf("expected stateChange on clear, got %+v", events[1])
+	}
+}
+
+func TestUpdateWithinSkewAccepted(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	// 5 seconds in the past is within the 30s Skew tolerance — accept it.
+	d := time.Now().Add(-5 * time.Second)
+	if err := w.Update(d); err != nil {
+		t.Fatalf("within-skew Update should succeed, got %v", err)
+	}
+	if !w.Deadline().Equal(d) {
+		t.Fatalf("expected deadline to be applied, got %v want %v", w.Deadline(), d)
+	}
+}
+
+func TestCloseSilencesUpdates(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	w.Close()
+
+	_ = w.Update(time.Now().Add(time.Hour))
+
+	time.Sleep(20 * time.Millisecond)
+	if got := r.snapshot(); len(got) != 0 {
+		t.Fatalf("expected no events after Close, got %+v", got)
+	}
+}
+
+// TestCloseClearsRecorderDeadline pins the profile-switch fix: a watcher
+// holding a live deadline must zero the recorder on Close so the next
+// engine's watcher (and the UI reading the shared server-scoped recorder)
+// doesn't start out showing the previous session's stale "expires in".
+func TestCloseClearsRecorderDeadline(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(time.Hour, r)
+
+	d := time.Now().Add(2 * time.Hour)
+	if err := w.Update(d); err != nil {
+		t.Fatalf("seed Update: %v", err)
+	}
+	if got := r.deadline(); !got.Equal(d) {
+		t.Fatalf("recorder deadline after Update = %v, want %v", got, d)
+	}
+
+	w.Close()
+
+	if got := r.deadline(); !got.IsZero() {
+		t.Fatalf("recorder deadline after Close = %v, want zero", got)
+	}
+}
+
+// TestCloseWithoutDeadlineLeavesRecorderUntouched guards the symmetric
+// case: closing a watcher that never held a deadline must not emit a
+// redundant clear (the recorder may legitimately hold a value written by
+// some other path; the watcher only owns what it set).
+func TestCloseWithoutDeadlineLeavesRecorderUntouched(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(time.Hour, r)
+
+	w.Close()
+
+	if got := r.snapshot(); len(got) != 0 {
+		t.Fatalf("expected no events from Close on an empty watcher, got %+v", got)
+	}
+}
+
+func TestFinalWarningFiresAfterRegularWarning(t *testing.T) {
+	r := &fakeRecorder{}
+	// Warning fires at deadline-80ms, final at deadline-30ms.
+	w := NewWithLeads(80*time.Millisecond, 30*time.Millisecond, r)
+	defer w.Close()
+
+	d := time.Now().Add(100 * time.Millisecond)
+	_ = w.Update(d)
+
+	// Expect stateChange + warning + final-warning.
+	events := waitForEvents(t, r, 3)
+
+	if countWhere(events, func(e event) bool { return e.kind == stateChange }) != 1 {
+		t.Fatalf("expected exactly 1 stateChange, got %+v", events)
+	}
+	if countWhere(events, event.isWarning) != 1 {
+		t.Fatalf("expected exactly 1 warning publish, got %+v", events)
+	}
+	if countWhere(events, event.isFinalWarning) != 1 {
+		t.Fatalf("expected exactly 1 final-warning publish, got %+v", events)
+	}
+
+	// Warning must precede final (same deadline, longer lead fires first).
+	var wIdx, fIdx int
+	for i, e := range events {
+		switch {
+		case e.isWarning():
+			wIdx = i
+		case e.isFinalWarning():
+			fIdx = i
+		}
+	}
+	if wIdx > fIdx {
+		t.Fatalf("warning must publish before final-warning, got order %+v", events)
+	}
+}
+
+func TestDismissSuppressesFinalWarning(t *testing.T) {
+	r := &fakeRecorder{}
+	w := NewWithLeads(80*time.Millisecond, 30*time.Millisecond, r)
+	defer w.Close()
+
+	d := time.Now().Add(100 * time.Millisecond)
+	_ = w.Update(d)
+
+	// Wait for the warning publish so we know we're inside the warning
+	// window, then dismiss before the final timer would fire.
+	deadline := time.Now().Add(500 * time.Millisecond)
+	for time.Now().Before(deadline) {
+		if countWhere(r.snapshot(), event.isWarning) >= 1 {
+			break
+		}
+		time.Sleep(2 * time.Millisecond)
+	}
+	if countWhere(r.snapshot(), event.isWarning) < 1 {
+		t.Fatalf("warning did not publish in time, events=%+v", r.snapshot())
+	}
+
+	w.Dismiss()
+
+	// Now wait past when the final would have fired.
+	time.Sleep(120 * time.Millisecond)
+
+	if n := countWhere(r.snapshot(), event.isFinalWarning); n != 0 {
+		t.Fatalf("final-warning published after Dismiss(), events=%+v", r.snapshot())
+	}
+}
+
+func TestDismissResetByNewDeadline(t *testing.T) {
+	r := &fakeRecorder{}
+	w := NewWithLeads(80*time.Millisecond, 30*time.Millisecond, r)
+	defer w.Close()
+
+	first := time.Now().Add(100 * time.Millisecond)
+	_ = w.Update(first)
+
+	// Dismiss against the first deadline.
+	w.Dismiss()
+
+	// Replace with a fresh deadline before the first's timers complete.
+	time.Sleep(10 * time.Millisecond)
+	second := time.Now().Add(100 * time.Millisecond)
+	_ = w.Update(second)
+
+	// The second cycle must publish a final-warning (the dismiss state
+	// did not carry over).
+	deadline := time.Now().Add(500 * time.Millisecond)
+	for time.Now().Before(deadline) {
+		if countWhere(r.snapshot(), event.isFinalWarning) >= 1 {
+			break
+		}
+		time.Sleep(5 * time.Millisecond)
+	}
+	if countWhere(r.snapshot(), event.isFinalWarning) < 1 {
+		t.Fatalf("final-warning did not publish on fresh deadline after Dismiss reset, events=%+v", r.snapshot())
+	}
+}
+
+func TestDismissBeforeUpdateIsNoop(t *testing.T) {
+	r := &fakeRecorder{}
+	w := NewWithLeads(80*time.Millisecond, 30*time.Millisecond, r)
+	defer w.Close()
+
+	// No deadline tracked yet; Dismiss must be a no-op (no panic, no state).
+	w.Dismiss()
+
+	d := time.Now().Add(100 * time.Millisecond)
+	_ = w.Update(d)
+
+	// Final warning should still publish — Dismiss only acts on the current
+	// deadline, and there was none at the time of the call.
+	deadline := time.Now().Add(500 * time.Millisecond)
+	for time.Now().Before(deadline) {
+		if countWhere(r.snapshot(), event.isFinalWarning) >= 1 {
+			return
+		}
+		time.Sleep(5 * time.Millisecond)
+	}
+	t.Fatalf("final-warning did not publish after no-op pre-Update Dismiss, events=%+v", r.snapshot())
+}

client/internal/conn_mgr.go

@@ -107,6 +107,37 @@ func (e *ConnMgr) UpdatedRemoteFeatureFlag(ctx context.Context, enabled bool) er
 	}
 }
 
+// SetLocalLazyConn applies a local lazy connection override (UI / CLI / env).
+// While enabledLocally is true, UpdatedRemoteFeatureFlag (management sync) is a
+// no-op, so the local setting wins until it is turned off again.
+func (e *ConnMgr) SetLocalLazyConn(ctx context.Context, enabled bool) error {
+	e.enabledLocally = enabled
+
+	if enabled {
+		if e.lazyConnMgr != nil {
+			return nil
+		}
+
+		if e.rosenpassEnabled {
+			log.Warnf("rosenpass connection manager is enabled, lazy connection manager will not be started")
+			return nil
+		}
+
+		log.Infof("lazy connection manager is enabled locally")
+		e.initLazyManager(ctx)
+		e.statusRecorder.UpdateLazyConnection(true)
+		return e.addPeersToLazyConnManager()
+	}
+
+	if e.lazyConnMgr == nil {
+		return nil
+	}
+	log.Infof("lazy connection manager is disabled locally")
+	e.closeManager(ctx)
+	e.statusRecorder.UpdateLazyConnection(false)
+	return nil
+}
+
 // UpdateRouteHAMap updates the route HA mappings in the lazy connection manager
 func (e *ConnMgr) UpdateRouteHAMap(haMap route.HAMap) {
 	if !e.isStartedWithLazyMgr() {

client/internal/connect.go

@@ -11,7 +11,6 @@ import (
 	"runtime/debug"
 	"strings"
 	"sync"
-	"sync/atomic"
 	"time"
 
 	"github.com/cenkalti/backoff/v4"
@@ -55,10 +54,6 @@ var androidRunOverride func(c *ConnectClient, runningChan chan struct{}, logPath
 
 type ConnectClient struct {
 	ctx            context.Context
-	runCancel      context.CancelFunc
-	runExited      chan struct{}
-	runOnce        sync.Once
-	runStarted     atomic.Bool
 	config         *profilemanager.Config
 	statusRecorder *peer.Status
 
@@ -75,14 +70,8 @@ func NewConnectClient(
 	config *profilemanager.Config,
 	statusRecorder *peer.Status,
 ) *ConnectClient {
-	// Derive the run context here so Stop owns the cancel that unblocks the run
-	// loop. runCancel is set once at construction, so Stop can call it without
-	// racing the run loop's startup. Callers therefore need not cancel before Stop.
-	runCtx, runCancel := context.WithCancel(ctx)
 	return &ConnectClient{
-		ctx:            runCtx,
-		runCancel:      runCancel,
-		runExited:      make(chan struct{}),
+		ctx:            ctx,
 		config:         config,
 		statusRecorder: statusRecorder,
 		engineMutex:    sync.Mutex{},
@@ -146,11 +135,6 @@ func (c *ConnectClient) RunOniOS(
 }
 
 func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan struct{}, logPath string) error {
-	// Mark the loop as started and signal exit on return so Stop can wait for
-	// the loop to finish (and skip the wait if the loop never ran).
-	c.runStarted.Store(true)
-	defer c.runOnce.Do(func() { close(c.runExited) })
-
 	defer func() {
 		if r := recover(); r != nil {
 			rec := c.statusRecorder
@@ -276,6 +260,15 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 		log.Debugf("connecting to the Management service %s", c.config.ManagementURL.Host)
 		mgmClient, err := mgm.NewClient(engineCtx, c.config.ManagementURL.Host, myPrivateKey, mgmTlsEnabled)
 		if err != nil {
+			// On daemon shutdown / Down() the parent context is cancelled
+			// and the dial fails with "context canceled". Wrapping that
+			// into state would leave the snapshot stuck at Connecting+err
+			// until the backoff loop wakes up — instead let the operation
+			// return cleanly so the deferred state.Set(StatusIdle) takes
+			// effect on the next iteration.
+			if c.ctx.Err() != nil {
+				return nil
+			}
 			return wrapErr(gstatus.Errorf(codes.FailedPrecondition, "failed connecting to Management Service : %s", err))
 		}
 		mgmNotifier := statusRecorderToMgmConnStateNotifier(c.statusRecorder)
@@ -306,7 +299,7 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 			log.Debug(err)
 			if s, ok := gstatus.FromError(err); ok && (s.Code() == codes.PermissionDenied) {
 				state.Set(StatusNeedsLogin)
-				c.runCancel()
+				_ = c.Stop()
 				return backoff.Permanent(wrapErr(err)) // unrecoverable error
 			}
 			return wrapErr(err)
@@ -409,6 +402,10 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 			return wrapErr(err)
 		}
 
+		// Seed the session-expiry deadline from the LoginResponse. Subsequent
+		// changes flow in through SyncResponse and are applied in handleSync.
+		engine.ApplySessionDeadline(loginResp.GetSessionExpiresAt())
+
 		log.Infof("Netbird engine started, the IP is: %s", peerConfig.GetAddress())
 		state.Set(StatusConnected)
 
@@ -426,10 +423,14 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 		c.engine = nil
 		c.engineMutex.Unlock()
 
-		log.Infof("ensuring wg interface is removed, Netbird engine context cancelled")
+		// todo: consider to remove this condition. Is not thread safe.
+		// We should always call Stop(), but we need to verify that it is idempotent
+		if engine.wgInterface != nil {
+			log.Infof("ensuring %s is removed, Netbird engine context cancelled", engine.wgInterface.Name())
 
-		if err := engine.Stop(); err != nil {
-			log.Errorf("Failed to stop engine: %v", err)
+			if err := engine.Stop(); err != nil {
+				log.Errorf("Failed to stop engine: %v", err)
+			}
 		}
 		c.statusRecorder.ClientTeardown()
 
@@ -445,12 +446,16 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 	}
 
 	c.statusRecorder.ClientStart()
+	// Wrap the backoff with c.ctx so Down()/actCancel propagates into the
+	// inter-attempt sleep — otherwise a 15s MaxInterval can keep the retry
+	// loop alive long after the caller asked to give up, leaving the
+	// status stream stuck at Connecting.
 	err = backoff.Retry(operation, backoff.WithContext(backOff, c.ctx))
 	if err != nil {
 		log.Debugf("exiting client retry loop due to unrecoverable error: %s", err)
 		if s, ok := gstatus.FromError(err); ok && (s.Code() == codes.PermissionDenied) {
 			state.Set(StatusNeedsLogin)
-			c.runCancel()
+			_ = c.Stop()
 		}
 		return err
 	}
@@ -528,9 +533,11 @@ func (c *ConnectClient) Status() StatusType {
 }
 
 func (c *ConnectClient) Stop() error {
-	c.runCancel()
-	if c.runStarted.Load() {
-		<-c.runExited
+	engine := c.Engine()
+	if engine != nil {
+		if err := engine.Stop(); err != nil {
+			return fmt.Errorf("stop engine: %w", err)
+		}
 	}
 	return nil
 }

client/internal/debug/debug.go

@@ -229,9 +229,16 @@ scutil_dns.txt (macOS only):
 
 const (
 	clientLogFile = "client.log"
+	uiLogFile     = "gui-client.log"
 	errorLogFile  = "netbird.err"
 	stdoutLogFile = "netbird.out"
 
+	// Rotated-log glob prefixes (base log name without extension) passed to
+	// addRotatedLogFiles. The daemon's own log and the GUI log live in the same
+	// dir, so the prefixes must be disjoint to keep their rotated siblings apart.
+	clientLogPrefix = "client"
+	uiLogPrefix     = "gui-client"
+
 	darwinErrorLogPath  = "/var/log/netbird.out.log"
 	darwinStdoutLogPath = "/var/log/netbird.err.log"
 )
@@ -249,6 +256,7 @@ type BundleGenerator struct {
 	statusRecorder *peer.Status
 	syncResponse   *mgmProto.SyncResponse
 	logPath        string
+	uiLogPath      string
 	tempDir        string
 	statePath      string
 	cpuProfile     []byte
@@ -276,6 +284,7 @@ type GeneratorDependencies struct {
 	StatusRecorder *peer.Status
 	SyncResponse   *mgmProto.SyncResponse
 	LogPath        string
+	UILogPath      string // Absolute path to the desktop UI's gui-client.log, reported via RegisterUILog. Empty if no UI registered one.
 	TempDir        string // Directory for temporary bundle zip files. If empty, os.TempDir() is used.
 	StatePath      string // Path to the state file. If empty, the ServiceManager default path is used.
 	CPUProfile     []byte
@@ -300,6 +309,7 @@ func NewBundleGenerator(deps GeneratorDependencies, cfg BundleConfig) *BundleGen
 		statusRecorder: deps.StatusRecorder,
 		syncResponse:   deps.SyncResponse,
 		logPath:        deps.LogPath,
+		uiLogPath:      deps.UILogPath,
 		tempDir:        deps.TempDir,
 		statePath:      deps.StatePath,
 		cpuProfile:     deps.CPUProfile,
@@ -411,6 +421,10 @@ func (g *BundleGenerator) createArchive() error {
 		log.Errorf("failed to add logs to debug bundle: %v", err)
 	}
 
+	if err := g.addUILog(); err != nil {
+		log.Errorf("failed to add UI log to debug bundle: %v", err)
+	}
+
 	if err := g.addUpdateLogs(); err != nil {
 		log.Errorf("failed to add updater logs: %v", err)
 	}
@@ -986,7 +1000,7 @@ func (g *BundleGenerator) addLogfile() error {
 		return fmt.Errorf("add client log file to zip: %w", err)
 	}
 
-	g.addRotatedLogFiles(logDir)
+	g.addRotatedLogFiles(logDir, clientLogPrefix)
 
 	stdErrLogPath := filepath.Join(logDir, errorLogFile)
 	stdoutLogPath := filepath.Join(logDir, stdoutLogFile)
@@ -1006,6 +1020,25 @@ func (g *BundleGenerator) addLogfile() error {
 	return nil
 }
 
+// addUILog adds the desktop UI's gui-client.log (and its rotated siblings) to
+// the bundle. The path is reported by the UI via RegisterUILog; empty when no
+// UI registered one (e.g. headless / server). Missing file is non-fatal — the
+// UI only writes it while the daemon is in debug, so it's often absent.
+func (g *BundleGenerator) addUILog() error {
+	if g.uiLogPath == "" {
+		log.Debugf("no UI log path registered, skipping in debug bundle")
+		return nil
+	}
+
+	if err := g.addSingleLogfile(g.uiLogPath, uiLogFile); err != nil {
+		return fmt.Errorf("add UI log file to zip: %w", err)
+	}
+
+	g.addRotatedLogFiles(filepath.Dir(g.uiLogPath), uiLogPrefix)
+
+	return nil
+}
+
 // addSingleLogfile adds a single log file to the archive
 func (g *BundleGenerator) addSingleLogfile(logPath, targetName string) error {
 	logFile, err := os.Open(logPath)
@@ -1078,14 +1111,16 @@ func (g *BundleGenerator) addSingleLogFileGz(logPath, targetName string) error {
 	return nil
 }
 
-// addRotatedLogFiles adds rotated log files to the bundle based on logFileCount
-func (g *BundleGenerator) addRotatedLogFiles(logDir string) {
+// addRotatedLogFiles adds rotated log files to the bundle based on logFileCount.
+// prefix is the base log name without extension (e.g. "client", "gui-client");
+// the glob matches both files rotated by us and by logrotate on linux.
+func (g *BundleGenerator) addRotatedLogFiles(logDir, prefix string) {
 	if g.logFileCount == 0 {
 		return
 	}
 
-	// This regex will match both logs rotated by us and logrotate on linux
-	pattern := filepath.Join(logDir, "client*.log.*")
+	// This pattern matches both logs rotated by us and logrotate on linux
+	pattern := filepath.Join(logDir, prefix+"*.log.*")
 	files, err := filepath.Glob(pattern)
 	if err != nil {
 		log.Warnf("failed to glob rotated logs: %v", err)

client/internal/debug/debug_logfiles_test.go

@@ -40,6 +40,25 @@ func TestAddRotatedLogFiles_PicksUpAllVariants(t *testing.T) {
 	require.NotContains(t, names, "other.log", "unrelated files should not be in bundle")
 }
 
+// TestAddRotatedLogFiles_GUIPrefix asserts the prefix parameter scopes the glob
+// to the GUI log: gui-client.log.* rotated siblings are picked up and the
+// daemon's own client.log.* are not (and vice versa, covered above). This is
+// the load-bearing check for the gui-client.log bundle collection — the old
+// "client*.log.*" glob would have missed gui-client rotations.
+func TestAddRotatedLogFiles_GUIPrefix(t *testing.T) {
+	dir := t.TempDir()
+
+	writeFile(t, filepath.Join(dir, "gui-client.log.1"), "gui rotated\n")
+	writeGzFile(t, filepath.Join(dir, "gui-client.log.2.gz"), "gui rotated gz\n")
+	writeFile(t, filepath.Join(dir, "client.log.1"), "daemon rotated\n")
+
+	names := runAddRotatedLogFilesPrefix(t, dir, "gui-client", 10)
+
+	require.Contains(t, names, "gui-client.log.1", "gui-client rotated file should be in bundle")
+	require.Contains(t, names, "gui-client.log.2.gz", "gui-client gz rotated file should be in bundle")
+	require.NotContains(t, names, "client.log.1", "daemon rotated file must not match the gui-client prefix")
+}
+
 // TestAddRotatedLogFiles_RespectsLogFileCount asserts that only the newest
 // logFileCount rotated files are bundled, ordered by mtime.
 func TestAddRotatedLogFiles_RespectsLogFileCount(t *testing.T) {
@@ -67,6 +86,10 @@ func TestAddRotatedLogFiles_RespectsLogFileCount(t *testing.T) {
 // runAddRotatedLogFiles calls addRotatedLogFiles against a fresh in-memory
 // zip writer and returns the set of entry names that ended up in the archive.
 func runAddRotatedLogFiles(t *testing.T, dir string, logFileCount uint32) map[string]struct{} {
+	return runAddRotatedLogFilesPrefix(t, dir, "client", logFileCount)
+}
+
+func runAddRotatedLogFilesPrefix(t *testing.T, dir, prefix string, logFileCount uint32) map[string]struct{} {
 	t.Helper()
 
 	var buf bytes.Buffer
@@ -74,7 +97,7 @@ func runAddRotatedLogFiles(t *testing.T, dir string, logFileCount uint32) map[st
 		archive:      zip.NewWriter(&buf),
 		logFileCount: logFileCount,
 	}
-	g.addRotatedLogFiles(dir)
+	g.addRotatedLogFiles(dir, prefix)
 	require.NoError(t, g.archive.Close())
 
 	zr, err := zip.NewReader(bytes.NewReader(buf.Bytes()), int64(buf.Len()))

client/internal/dns/mgmt/mgmt.go

@@ -51,20 +51,13 @@ type cachedRecord struct {
 }
 
 // Resolver caches critical NetBird infrastructure domains.
-// records, refreshing, failedResolves, mgmtDomain and serverDomains are all
-// guarded by mutex.
+// records, refreshing, mgmtDomain and serverDomains are all guarded by mutex.
 type Resolver struct {
 	records       map[dns.Question]*cachedRecord
 	mgmtDomain    *domain.Domain
 	serverDomains *dnsconfig.ServerDomains
 	mutex         sync.RWMutex
 
-	// failedResolves records the last failed initial resolve per domain so a
-	// domain that never resolves isn't retried on every server-domains update
-	// until refreshBackoff elapses. Entries are cleared on success and pruned
-	// to the current server-domains set.
-	failedResolves map[domain.Domain]time.Time
-
 	chain            ChainResolver
 	chainMaxPriority int
 	refreshGroup     singleflight.Group
@@ -83,10 +76,9 @@ type Resolver struct {
 // NewResolver creates a new management domains cache resolver.
 func NewResolver() *Resolver {
 	return &Resolver{
-		records:        make(map[dns.Question]*cachedRecord),
-		refreshing:     make(map[dns.Question]*atomic.Bool),
-		failedResolves: make(map[domain.Domain]time.Time),
-		cacheTTL:       resolveCacheTTL(),
+		records:    make(map[dns.Question]*cachedRecord),
+		refreshing: make(map[dns.Question]*atomic.Bool),
+		cacheTTL:   resolveCacheTTL(),
 	}
 }
 
@@ -181,9 +173,7 @@ func (m *Resolver) continueToNext(w dns.ResponseWriter, r *dns.Msg) {
 
 // AddDomain resolves a domain and stores its A/AAAA records in the cache.
 // A family that resolves NODATA (nil err, zero records) evicts any stale
-// entry for that qtype. When one family hard-errors while the other succeeds,
-// the resolved family is still cached but AddDomain returns an error so the
-// caller retries the incomplete resolve rather than treating it as complete.
+// entry for that qtype.
 func (m *Resolver) AddDomain(ctx context.Context, d domain.Domain) error {
 	dnsName := strings.ToLower(dns.Fqdn(d.PunycodeString()))
 
@@ -213,10 +203,6 @@ func (m *Resolver) AddDomain(ctx context.Context, d domain.Domain) error {
 	log.Debugf("added/updated domain=%s with %d A records and %d AAAA records",
 		d.SafeString(), len(aRecords), len(aaaaRecords))
 
-	if errA != nil || errAAAA != nil {
-		return fmt.Errorf("resolve %s: incomplete, a family failed: %w", d.SafeString(), errors.Join(errA, errAAAA))
-	}
-
 	return nil
 }
 
@@ -476,7 +462,6 @@ func (m *Resolver) RemoveDomain(d domain.Domain) error {
 	delete(m.records, qAAAA)
 	delete(m.refreshing, qA)
 	delete(m.refreshing, qAAAA)
-	delete(m.failedResolves, d)
 
 	log.Debugf("removed domain=%s from cache", d.SafeString())
 	return nil
@@ -520,7 +505,6 @@ func (m *Resolver) UpdateFromServerDomains(ctx context.Context, serverDomains dn
 		allDomains := m.extractDomainsFromServerDomains(updatedServerDomains)
 		currentDomains := m.GetCachedDomains()
 		removedDomains = m.removeStaleDomains(currentDomains, allDomains)
-		m.pruneFailedResolves(allDomains)
 	}
 
 	m.addNewDomains(ctx, newDomains)
@@ -593,85 +577,13 @@ func (m *Resolver) isManagementDomain(domain domain.Domain) bool {
 	return m.mgmtDomain != nil && domain == *m.mgmtDomain
 }
 
-// addNewDomains resolves and caches domains that are not yet in the cache,
-// running the lookups concurrently. Domains already cached are skipped and left
-// to the stale-while-revalidate refresh path, so a sync never re-resolves them
-// synchronously: once NetBird owns the OS resolver the resolve runs through the
-// handler chain and would otherwise dial the managed upstreams under the engine
-// sync lock on every update.
+// addNewDomains resolves and caches all domains from the update
 func (m *Resolver) addNewDomains(ctx context.Context, newDomains domain.List) {
-	var wg sync.WaitGroup
-	seen := make(map[domain.Domain]struct{}, len(newDomains))
 	for _, newDomain := range newDomains {
-		if _, dup := seen[newDomain]; dup {
-			continue
-		}
-		seen[newDomain] = struct{}{}
-
-		if !m.needsResolve(newDomain) {
-			continue
-		}
-
-		wg.Add(1)
-		go func(d domain.Domain) {
-			defer wg.Done()
-			if err := m.AddDomain(ctx, d); err != nil {
-				m.markResolveFailed(d)
-				log.Warnf("failed to add/update domain=%s: %v", d.SafeString(), err)
-				return
-			}
-			m.clearResolveFailed(d)
-			log.Debugf("added/updated management cache domain=%s", d.SafeString())
-		}(newDomain)
-	}
-	wg.Wait()
-}
-
-// needsResolve reports whether d should be resolved now. A recent failed or
-// incomplete resolve gates retries on the backoff even when one family is
-// already cached, so a transiently-failed family is retried instead of being
-// treated as fully resolved. Otherwise a domain with any cached record is left
-// to the stale-while-revalidate refresh path.
-func (m *Resolver) needsResolve(d domain.Domain) bool {
-	dnsName := strings.ToLower(dns.Fqdn(d.PunycodeString()))
-
-	m.mutex.RLock()
-	defer m.mutex.RUnlock()
-
-	if failedAt, ok := m.failedResolves[d]; ok {
-		return time.Since(failedAt) >= refreshBackoff
-	}
-
-	for _, qtype := range []uint16{dns.TypeA, dns.TypeAAAA} {
-		q := dns.Question{Name: dnsName, Qtype: qtype, Qclass: dns.ClassINET}
-		if _, ok := m.records[q]; ok {
-			return false
-		}
-	}
-	return true
-}
-
-func (m *Resolver) markResolveFailed(d domain.Domain) {
-	m.mutex.Lock()
-	m.failedResolves[d] = time.Now()
-	m.mutex.Unlock()
-}
-
-func (m *Resolver) clearResolveFailed(d domain.Domain) {
-	m.mutex.Lock()
-	delete(m.failedResolves, d)
-	m.mutex.Unlock()
-}
-
-// pruneFailedResolves drops failure markers for domains no longer present in
-// the server-domains set, keeping the map bounded to the current set (a
-// failed-only domain has no cached record, so RemoveDomain never sees it).
-func (m *Resolver) pruneFailedResolves(domains domain.List) {
-	m.mutex.Lock()
-	defer m.mutex.Unlock()
-	for d := range m.failedResolves {
-		if !slices.Contains(domains, d) {
-			delete(m.failedResolves, d)
+		if err := m.AddDomain(ctx, newDomain); err != nil {
+			log.Warnf("failed to add/update domain=%s: %v", newDomain.SafeString(), err)
+		} else {
+			log.Debugf("added/updated management cache domain=%s", newDomain.SafeString())
 		}
 	}
 }

client/internal/dns/mgmt/mgmt_refresh_test.go

@@ -21,7 +21,6 @@ type fakeChain struct {
 	mu       sync.Mutex
 	calls    map[string]int
 	answers  map[string][]dns.RR
-	qErr     map[string]error
 	err      error
 	hasRoot  bool
 	onLookup func()
@@ -31,7 +30,6 @@ func newFakeChain() *fakeChain {
 	return &fakeChain{
 		calls:   map[string]int{},
 		answers: map[string][]dns.RR{},
-		qErr:    map[string]error{},
 		hasRoot: true,
 	}
 }
@@ -49,9 +47,6 @@ func (f *fakeChain) ResolveInternal(ctx context.Context, msg *dns.Msg, maxPriori
 	f.calls[key]++
 	answers := f.answers[key]
 	err := f.err
-	if err == nil {
-		err = f.qErr[key]
-	}
 	onLookup := f.onLookup
 	f.mu.Unlock()
 
@@ -80,12 +75,6 @@ func (f *fakeChain) setAnswer(name string, qtype uint16, ip string) {
 	}
 }
 
-func (f *fakeChain) setErr(name string, qtype uint16, err error) {
-	f.mu.Lock()
-	defer f.mu.Unlock()
-	f.qErr[name+"|"+dns.TypeToString[qtype]] = err
-}
-
 func (f *fakeChain) callCount(name string, qtype uint16) int {
 	f.mu.Lock()
 	defer f.mu.Unlock()

client/internal/dns/mgmt/mgmt_resolve_test.go

@@ -1,183 +0,0 @@
-package mgmt
-
-import (
-	"context"
-	"errors"
-	"sync/atomic"
-	"testing"
-	"time"
-
-	"github.com/miekg/dns"
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	dnsconfig "github.com/netbirdio/netbird/client/internal/dns/config"
-	"github.com/netbirdio/netbird/shared/management/domain"
-)
-
-// A domain already in the cache must not be re-resolved on a subsequent server
-// domains update; it is left to the stale-while-revalidate refresh path.
-func TestResolver_UpdateFromServerDomains_SkipsCached(t *testing.T) {
-	r := NewResolver()
-	chain := newFakeChain()
-	chain.setAnswer("signal.example.com.", dns.TypeA, "10.0.0.2")
-	r.SetChainResolver(chain, 50)
-
-	sd := dnsconfig.ServerDomains{Signal: domain.Domain("signal.example.com")}
-
-	_, err := r.UpdateFromServerDomains(context.Background(), sd)
-	require.NoError(t, err)
-	require.Equal(t, 1, chain.callCount("signal.example.com.", dns.TypeA),
-		"first update must resolve the domain")
-
-	_, err = r.UpdateFromServerDomains(context.Background(), sd)
-	require.NoError(t, err)
-	assert.Equal(t, 1, chain.callCount("signal.example.com.", dns.TypeA),
-		"cached domain must not be re-resolved on a subsequent update")
-}
-
-// New domains in a single update must resolve concurrently rather than serially.
-func TestResolver_AddNewDomains_ResolvesConcurrently(t *testing.T) {
-	r := NewResolver()
-	chain := newFakeChain()
-
-	var inflight, maxInflight atomic.Int32
-	chain.onLookup = func() {
-		n := inflight.Add(1)
-		for {
-			old := maxInflight.Load()
-			if n <= old || maxInflight.CompareAndSwap(old, n) {
-				break
-			}
-		}
-		time.Sleep(50 * time.Millisecond)
-		inflight.Add(-1)
-	}
-
-	relays := []domain.Domain{"a.example.com", "b.example.com", "c.example.com", "d.example.com"}
-	for _, d := range relays {
-		chain.setAnswer(dns.Fqdn(string(d)), dns.TypeA, "10.0.0.2")
-	}
-	r.SetChainResolver(chain, 50)
-
-	start := time.Now()
-	_, err := r.UpdateFromServerDomains(context.Background(), dnsconfig.ServerDomains{Relay: relays})
-	require.NoError(t, err)
-	elapsed := time.Since(start)
-
-	assert.GreaterOrEqual(t, int(maxInflight.Load()), 2, "domains must resolve concurrently")
-	// Serial resolution of 4 domains would take at least 4*50ms; concurrent is far less.
-	assert.Less(t, elapsed, 300*time.Millisecond, "resolution should not be serial")
-}
-
-// A domain that fails to resolve must not be retried on every update; the
-// failure backoff suppresses re-resolution until it expires.
-func TestResolver_UpdateFromServerDomains_BacksOffFailures(t *testing.T) {
-	r := NewResolver()
-	chain := newFakeChain()
-	chain.err = errors.New("resolve boom")
-	r.SetChainResolver(chain, 50)
-
-	sd := dnsconfig.ServerDomains{Signal: domain.Domain("signal.example.com")}
-
-	_, err := r.UpdateFromServerDomains(context.Background(), sd)
-	require.NoError(t, err)
-	require.Equal(t, 1, chain.callCount("signal.example.com.", dns.TypeA),
-		"first update must attempt the resolve")
-
-	_, err = r.UpdateFromServerDomains(context.Background(), sd)
-	require.NoError(t, err)
-	assert.Equal(t, 1, chain.callCount("signal.example.com.", dns.TypeA),
-		"failed resolve must back off and not retry on the next update")
-}
-
-// A domain listed under more than one server-domain type (e.g. STUN and TURN on
-// the same host) must be resolved once per update, not once per occurrence.
-func TestResolver_AddNewDomains_DedupesDuplicateDomains(t *testing.T) {
-	r := NewResolver()
-	chain := newFakeChain()
-	chain.setAnswer("dup.example.com.", dns.TypeA, "10.0.0.9")
-	r.SetChainResolver(chain, 50)
-
-	sd := dnsconfig.ServerDomains{
-		Stuns: []domain.Domain{"dup.example.com"},
-		Turns: []domain.Domain{"dup.example.com"},
-	}
-
-	_, err := r.UpdateFromServerDomains(context.Background(), sd)
-	require.NoError(t, err)
-	assert.Equal(t, 1, chain.callCount("dup.example.com.", dns.TypeA),
-		"a domain appearing under multiple server-domain types must resolve once")
-}
-
-// A failure marker must be dropped once its domain leaves the server-domains set
-// so the map stays bounded to the current set.
-func TestResolver_UpdateFromServerDomains_PrunesFailedResolves(t *testing.T) {
-	r := NewResolver()
-	chain := newFakeChain()
-	chain.err = errors.New("resolve boom")
-	r.SetChainResolver(chain, 50)
-
-	_, err := r.UpdateFromServerDomains(context.Background(), dnsconfig.ServerDomains{Signal: domain.Domain("gone.example.com")})
-	require.NoError(t, err)
-	r.mutex.RLock()
-	_, marked := r.failedResolves[domain.Domain("gone.example.com")]
-	r.mutex.RUnlock()
-	require.True(t, marked, "failed resolve must be recorded")
-
-	_, err = r.UpdateFromServerDomains(context.Background(), dnsconfig.ServerDomains{Signal: domain.Domain("other.example.com")})
-	require.NoError(t, err)
-	r.mutex.RLock()
-	_, stillMarked := r.failedResolves[domain.Domain("gone.example.com")]
-	r.mutex.RUnlock()
-	assert.False(t, stillMarked, "failure marker for a domain no longer in the set must be pruned")
-}
-
-// When one family hard-errors while the other resolves, the domain is cached
-// for the working family but recorded as incomplete so the failed family is
-// retried under backoff instead of being treated as fully resolved forever.
-func TestResolver_AddNewDomains_RetriesPartialFamilyFailure(t *testing.T) {
-	d := domain.Domain("relay.example.com")
-	r := NewResolver()
-	chain := newFakeChain()
-	chain.setAnswer("relay.example.com.", dns.TypeA, "10.0.0.2")
-	chain.setErr("relay.example.com.", dns.TypeAAAA, errors.New("servfail"))
-	r.SetChainResolver(chain, 50)
-
-	_, err := r.UpdateFromServerDomains(context.Background(), dnsconfig.ServerDomains{Relay: []domain.Domain{d}})
-	require.NoError(t, err)
-
-	r.mutex.RLock()
-	_, aCached := r.records[dns.Question{Name: "relay.example.com.", Qtype: dns.TypeA, Qclass: dns.ClassINET}]
-	_, marked := r.failedResolves[d]
-	r.mutex.RUnlock()
-	require.True(t, aCached, "the working family must still be cached")
-	require.True(t, marked, "a partial failure must be recorded so the failed family is retried")
-
-	assert.False(t, r.needsResolve(d), "within the backoff window the domain is not retried")
-
-	r.mutex.Lock()
-	r.failedResolves[d] = time.Now().Add(-2 * refreshBackoff)
-	r.mutex.Unlock()
-	assert.True(t, r.needsResolve(d), "after the backoff elapses the domain is retried to pick up the missing family")
-}
-
-// A family that returns NODATA (legitimately absent, e.g. an IPv4-only host) is
-// not a failure: the domain must not be marked for retry, otherwise it would be
-// re-resolved on every sync.
-func TestResolver_AddNewDomains_NodataIsNotFailure(t *testing.T) {
-	d := domain.Domain("v4only.example.com")
-	r := NewResolver()
-	chain := newFakeChain()
-	chain.setAnswer("v4only.example.com.", dns.TypeA, "10.0.0.2")
-	r.SetChainResolver(chain, 50)
-
-	_, err := r.UpdateFromServerDomains(context.Background(), dnsconfig.ServerDomains{Relay: []domain.Domain{d}})
-	require.NoError(t, err)
-
-	r.mutex.RLock()
-	_, marked := r.failedResolves[d]
-	r.mutex.RUnlock()
-	assert.False(t, marked, "a NODATA family must not be recorded as a failure")
-	assert.False(t, r.needsResolve(d), "an IPv4-only host must not be re-resolved on later syncs")
-}

client/internal/dns/resutil/resolve.go

@@ -8,7 +8,6 @@ import (
 	"errors"
 	"net"
 	"net/netip"
-	"slices"
 	"strings"
 
 	"github.com/miekg/dns"
@@ -168,10 +167,7 @@ func getRcodeForNotFound(ctx context.Context, r resolver, domain string, origina
 	case dns.TypeA:
 		alternativeNetwork = "ip6"
 	default:
-		// Non-address types reach LookupIP only unexpectedly; without an
-		// address pair to probe we cannot prove the name is absent, so answer
-		// NODATA rather than a poisoning NXDOMAIN.
-		return dns.RcodeSuccess
+		return dns.RcodeNameError
 	}
 
 	if _, err := r.LookupNetIP(ctx, alternativeNetwork, domain); err != nil {
@@ -188,230 +184,6 @@ func getRcodeForNotFound(ctx context.Context, r resolver, domain string, origina
 	return dns.RcodeSuccess
 }
 
-// RecordResolver is the host resolver surface used to forward non-address
-// record queries. net.DefaultResolver satisfies it.
-type RecordResolver interface {
-	LookupMX(ctx context.Context, name string) ([]*net.MX, error)
-	LookupTXT(ctx context.Context, name string) ([]string, error)
-	LookupNS(ctx context.Context, name string) ([]*net.NS, error)
-	LookupSRV(ctx context.Context, service, proto, name string) (string, []*net.SRV, error)
-	LookupCNAME(ctx context.Context, host string) (string, error)
-	LookupAddr(ctx context.Context, addr string) ([]string, error)
-}
-
-// LookupRecords resolves a non-address DNS record type through the host
-// resolver and returns the resource records and the DNS rcode. Types the host
-// resolver cannot answer (anything not covered by the net.Resolver Lookup*
-// methods) yield NODATA so that a routed name is never poisoned with NXDOMAIN
-// for an unsupported type.
-func LookupRecords(ctx context.Context, r RecordResolver, name string, qtype uint16, ttl uint32) ([]dns.RR, int) {
-	fqdn := dns.Fqdn(name)
-
-	switch qtype {
-	case dns.TypeMX:
-		return lookupMX(ctx, r, name, fqdn, ttl)
-	case dns.TypeTXT:
-		return lookupTXT(ctx, r, name, fqdn, ttl)
-	case dns.TypeNS:
-		return lookupNS(ctx, r, name, fqdn, ttl)
-	case dns.TypeSRV:
-		return lookupSRV(ctx, r, name, fqdn, ttl)
-	case dns.TypeCNAME:
-		return lookupCNAME(ctx, r, name, fqdn, ttl)
-	case dns.TypePTR:
-		return lookupPTR(ctx, r, name, fqdn, ttl)
-	default:
-		return nil, dns.RcodeSuccess
-	}
-}
-
-func recordHeader(fqdn string, rrtype uint16, ttl uint32) dns.RR_Header {
-	return dns.RR_Header{Name: fqdn, Rrtype: rrtype, Class: dns.ClassINET, Ttl: ttl}
-}
-
-func lookupMX(ctx context.Context, r RecordResolver, name, fqdn string, ttl uint32) ([]dns.RR, int) {
-	recs, err := r.LookupMX(ctx, name)
-	if err != nil {
-		return nil, rcodeForRecordError(err)
-	}
-	rrs := make([]dns.RR, 0, len(recs))
-	for _, mx := range recs {
-		rrs = append(rrs, &dns.MX{
-			Hdr:        recordHeader(fqdn, dns.TypeMX, ttl),
-			Preference: mx.Pref,
-			Mx:         dns.Fqdn(mx.Host),
-		})
-	}
-	return rrs, dns.RcodeSuccess
-}
-
-func lookupTXT(ctx context.Context, r RecordResolver, name, fqdn string, ttl uint32) ([]dns.RR, int) {
-	recs, err := r.LookupTXT(ctx, name)
-	if err != nil {
-		return nil, rcodeForRecordError(err)
-	}
-	rrs := make([]dns.RR, 0, len(recs))
-	for _, txt := range recs {
-		rrs = append(rrs, &dns.TXT{
-			Hdr: recordHeader(fqdn, dns.TypeTXT, ttl),
-			Txt: chunkTXT(txt),
-		})
-	}
-	return rrs, dns.RcodeSuccess
-}
-
-func lookupNS(ctx context.Context, r RecordResolver, name, fqdn string, ttl uint32) ([]dns.RR, int) {
-	recs, err := r.LookupNS(ctx, name)
-	if err != nil {
-		return nil, rcodeForRecordError(err)
-	}
-	rrs := make([]dns.RR, 0, len(recs))
-	for _, ns := range recs {
-		rrs = append(rrs, &dns.NS{
-			Hdr: recordHeader(fqdn, dns.TypeNS, ttl),
-			Ns:  dns.Fqdn(ns.Host),
-		})
-	}
-	return rrs, dns.RcodeSuccess
-}
-
-func lookupSRV(ctx context.Context, r RecordResolver, name, fqdn string, ttl uint32) ([]dns.RR, int) {
-	_, recs, err := r.LookupSRV(ctx, "", "", name)
-	if err != nil {
-		return nil, rcodeForRecordError(err)
-	}
-	rrs := make([]dns.RR, 0, len(recs))
-	for _, srv := range recs {
-		rrs = append(rrs, &dns.SRV{
-			Hdr:      recordHeader(fqdn, dns.TypeSRV, ttl),
-			Priority: srv.Priority,
-			Weight:   srv.Weight,
-			Port:     srv.Port,
-			Target:   dns.Fqdn(srv.Target),
-		})
-	}
-	return rrs, dns.RcodeSuccess
-}
-
-func lookupCNAME(ctx context.Context, r RecordResolver, name, fqdn string, ttl uint32) ([]dns.RR, int) {
-	cname, err := r.LookupCNAME(ctx, name)
-	if err != nil {
-		return nil, rcodeForRecordError(err)
-	}
-	// LookupCNAME returns the queried name itself when the name resolves but
-	// has no CNAME record; that is a NODATA result, not a CNAME.
-	if strings.EqualFold(dns.Fqdn(cname), fqdn) {
-		return nil, dns.RcodeSuccess
-	}
-	return []dns.RR{&dns.CNAME{
-		Hdr:    recordHeader(fqdn, dns.TypeCNAME, ttl),
-		Target: dns.Fqdn(cname),
-	}}, dns.RcodeSuccess
-}
-
-func lookupPTR(ctx context.Context, r RecordResolver, name, fqdn string, ttl uint32) ([]dns.RR, int) {
-	addr, ok := ptrQueryAddr(name)
-	if !ok {
-		return nil, dns.RcodeSuccess
-	}
-	names, err := r.LookupAddr(ctx, addr)
-	if err != nil {
-		return nil, rcodeForRecordError(err)
-	}
-	rrs := make([]dns.RR, 0, len(names))
-	for _, n := range names {
-		rrs = append(rrs, &dns.PTR{
-			Hdr: recordHeader(fqdn, dns.TypePTR, ttl),
-			Ptr: dns.Fqdn(n),
-		})
-	}
-	return rrs, dns.RcodeSuccess
-}
-
-// ptrQueryAddr converts a reverse-DNS query name (in-addr.arpa or ip6.arpa)
-// into the address string expected by net.Resolver.LookupAddr. It reports false
-// when the name is not a well-formed reverse name.
-func ptrQueryAddr(qname string) (string, bool) {
-	name := strings.TrimSuffix(strings.ToLower(dns.Fqdn(qname)), ".")
-
-	switch {
-	case strings.HasSuffix(name, ".in-addr.arpa"):
-		return parseInAddrArpa(strings.TrimSuffix(name, ".in-addr.arpa"))
-	case strings.HasSuffix(name, ".ip6.arpa"):
-		return parseIP6Arpa(strings.TrimSuffix(name, ".ip6.arpa"))
-	default:
-		return "", false
-	}
-}
-
-// parseInAddrArpa turns the label portion of an in-addr.arpa name into an IPv4
-// address string, reporting false when it is not a well-formed reverse name.
-func parseInAddrArpa(labelPart string) (string, bool) {
-	labels := strings.Split(labelPart, ".")
-	if len(labels) != 4 {
-		return "", false
-	}
-	slices.Reverse(labels)
-	addr, err := netip.ParseAddr(strings.Join(labels, "."))
-	if err != nil || !addr.Is4() {
-		return "", false
-	}
-	return addr.String(), true
-}
-
-// parseIP6Arpa turns the nibble portion of an ip6.arpa name into an IPv6
-// address string, reporting false when it is not a well-formed reverse name.
-func parseIP6Arpa(nibblePart string) (string, bool) {
-	nibbles := strings.Split(nibblePart, ".")
-	if len(nibbles) != 32 {
-		return "", false
-	}
-	slices.Reverse(nibbles)
-	var sb strings.Builder
-	for i, n := range nibbles {
-		if i > 0 && i%4 == 0 {
-			sb.WriteByte(':')
-		}
-		sb.WriteString(n)
-	}
-	addr, err := netip.ParseAddr(sb.String())
-	if err != nil || !addr.Is6() {
-		return "", false
-	}
-	return addr.String(), true
-}
-
-// rcodeForRecordError maps a non-address lookup error to a DNS rcode. A
-// not-found result becomes NODATA rather than NXDOMAIN: net.DNSError.IsNotFound
-// does not distinguish a missing name from a name that exists only with records
-// of other types, so the name cannot be proven absent and must not be poisoned.
-func rcodeForRecordError(err error) int {
-	var dnsErr *net.DNSError
-	if errors.As(err, &dnsErr) && dnsErr.IsNotFound {
-		return dns.RcodeSuccess
-	}
-	return dns.RcodeServerFailure
-}
-
-// chunkTXT splits a TXT string into character-strings no longer than 255 bytes
-// so the record can be packed. The chunks form one TXT resource record.
-func chunkTXT(s string) []string {
-	const maxLen = 255
-	if len(s) <= maxLen {
-		return []string{s}
-	}
-
-	var chunks []string
-	for len(s) > maxLen {
-		chunks = append(chunks, s[:maxLen])
-		s = s[maxLen:]
-	}
-	if len(s) > 0 {
-		chunks = append(chunks, s)
-	}
-	return chunks
-}
-
 // FormatAnswers formats DNS resource records for logging.
 func FormatAnswers(answers []dns.RR) string {
 	if len(answers) == 0 {
@@ -435,35 +207,3 @@ func FormatAnswers(answers []dns.RR) string {
 	}
 	return "[" + strings.Join(parts, ", ") + "]"
 }
-
-// StripOPT removes any OPT pseudo-RRs from the message's Extra section. Per
-// RFC 6891 a responder must not include an OPT RR toward a client that did not
-// advertise EDNS0.
-func StripOPT(msg *dns.Msg) {
-	if len(msg.Extra) == 0 {
-		return
-	}
-	out := msg.Extra[:0]
-	for _, rr := range msg.Extra {
-		if _, ok := rr.(*dns.OPT); ok {
-			continue
-		}
-		out = append(out, rr)
-	}
-	msg.Extra = out
-}
-
-// ExtractEDE returns the first Extended DNS Error (RFC 8914) option carried in
-// the message, if present.
-func ExtractEDE(msg *dns.Msg) (*dns.EDNS0_EDE, bool) {
-	opt := msg.IsEdns0()
-	if opt == nil {
-		return nil, false
-	}
-	for _, o := range opt.Option {
-		if ede, ok := o.(*dns.EDNS0_EDE); ok {
-			return ede, true
-		}
-	}
-	return nil, false
-}

client/internal/dns/resutil/resolve_test.go

@@ -5,7 +5,6 @@ import (
 	"errors"
 	"net"
 	"net/netip"
-	"strings"
 	"testing"
 
 	"github.com/miekg/dns"
@@ -121,200 +120,3 @@ func TestLookupIP_DNSErrorNotIsNotFound(t *testing.T) {
 
 	assert.Equal(t, dns.RcodeServerFailure, result.Rcode, "upstream failure should map to SERVFAIL")
 }
-
-func TestPtrQueryAddr(t *testing.T) {
-	tests := []struct {
-		name   string
-		qname  string
-		want   string
-		wantOK bool
-	}{
-		{name: "ipv4", qname: "4.3.2.1.in-addr.arpa.", want: "1.2.3.4", wantOK: true},
-		{name: "ipv4 no trailing dot", qname: "1.0.0.127.in-addr.arpa", want: "127.0.0.1", wantOK: true},
-		{
-			name:   "ipv6",
-			qname:  "1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.8.b.d.0.1.0.0.2.ip6.arpa.",
-			want:   "2001:db8::1",
-			wantOK: true,
-		},
-		{name: "ipv4 wrong label count", qname: "2.1.in-addr.arpa.", wantOK: false},
-		{name: "ipv6 wrong nibble count", qname: "1.0.ip6.arpa.", wantOK: false},
-		{name: "not a reverse name", qname: "example.com.", wantOK: false},
-		{name: "ipv4 bad octet", qname: "4.3.2.999.in-addr.arpa.", wantOK: false},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			got, ok := ptrQueryAddr(tt.qname)
-			assert.Equal(t, tt.wantOK, ok, "parse success mismatch")
-			if tt.wantOK {
-				assert.Equal(t, tt.want, got, "parsed address mismatch")
-			}
-		})
-	}
-}
-
-type mockRecordResolver struct {
-	mx    []*net.MX
-	txt   []string
-	ns    []*net.NS
-	srv   []*net.SRV
-	cname string
-	ptr   []string
-	err   error
-}
-
-func (m *mockRecordResolver) LookupMX(context.Context, string) ([]*net.MX, error) {
-	return m.mx, m.err
-}
-func (m *mockRecordResolver) LookupTXT(context.Context, string) ([]string, error) {
-	return m.txt, m.err
-}
-func (m *mockRecordResolver) LookupNS(context.Context, string) ([]*net.NS, error) {
-	return m.ns, m.err
-}
-func (m *mockRecordResolver) LookupSRV(context.Context, string, string, string) (string, []*net.SRV, error) {
-	return "", m.srv, m.err
-}
-func (m *mockRecordResolver) LookupCNAME(context.Context, string) (string, error) {
-	return m.cname, m.err
-}
-func (m *mockRecordResolver) LookupAddr(context.Context, string) ([]string, error) {
-	return m.ptr, m.err
-}
-
-func TestLookupRecords(t *testing.T) {
-	notFound := &net.DNSError{IsNotFound: true, Name: "example.com."}
-
-	t.Run("MX success", func(t *testing.T) {
-		r := &mockRecordResolver{mx: []*net.MX{{Host: "mail.example.com.", Pref: 10}}}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeMX, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		assert.Equal(t, "mail.example.com.", rrs[0].(*dns.MX).Mx)
-	})
-
-	t.Run("TXT short string is one character-string", func(t *testing.T) {
-		r := &mockRecordResolver{txt: []string{"v=spf1 -all"}}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeTXT, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		assert.Equal(t, []string{"v=spf1 -all"}, rrs[0].(*dns.TXT).Txt)
-	})
-
-	t.Run("TXT chunks long strings", func(t *testing.T) {
-		long := strings.Repeat("a", 300)
-		r := &mockRecordResolver{txt: []string{long}}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeTXT, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		txt := rrs[0].(*dns.TXT).Txt
-		require.Len(t, txt, 2, "300-byte string should split into two character-strings")
-		assert.Equal(t, 255, len(txt[0]))
-		assert.Equal(t, 45, len(txt[1]))
-	})
-
-	t.Run("NS success", func(t *testing.T) {
-		r := &mockRecordResolver{ns: []*net.NS{{Host: "ns1.example.com."}}}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeNS, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		assert.Equal(t, "ns1.example.com.", rrs[0].(*dns.NS).Ns)
-	})
-
-	t.Run("SRV success", func(t *testing.T) {
-		r := &mockRecordResolver{srv: []*net.SRV{{Target: "sip.example.com.", Port: 5060}}}
-		rrs, rcode := LookupRecords(context.Background(), r, "_sip._tcp.example.com.", dns.TypeSRV, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		assert.Equal(t, uint16(5060), rrs[0].(*dns.SRV).Port)
-	})
-
-	t.Run("CNAME success", func(t *testing.T) {
-		r := &mockRecordResolver{cname: "target.example.com."}
-		rrs, rcode := LookupRecords(context.Background(), r, "www.example.com.", dns.TypeCNAME, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		assert.Equal(t, "target.example.com.", rrs[0].(*dns.CNAME).Target)
-	})
-
-	t.Run("CNAME equal to name is NODATA", func(t *testing.T) {
-		r := &mockRecordResolver{cname: "example.com."}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeCNAME, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		assert.Empty(t, rrs, "self-referential CNAME is NODATA")
-	})
-
-	t.Run("PTR success", func(t *testing.T) {
-		r := &mockRecordResolver{ptr: []string{"host.example.com."}}
-		rrs, rcode := LookupRecords(context.Background(), r, "4.3.2.1.in-addr.arpa.", dns.TypePTR, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		require.Len(t, rrs, 1)
-		assert.Equal(t, "host.example.com.", rrs[0].(*dns.PTR).Ptr)
-	})
-
-	t.Run("PTR malformed name is NODATA", func(t *testing.T) {
-		r := &mockRecordResolver{}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypePTR, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		assert.Empty(t, rrs)
-	})
-
-	t.Run("not found is NODATA never NXDOMAIN", func(t *testing.T) {
-		r := &mockRecordResolver{err: notFound}
-		_, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeMX, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode, "missing record must not poison the name")
-	})
-
-	t.Run("server failure maps to SERVFAIL", func(t *testing.T) {
-		r := &mockRecordResolver{err: &net.DNSError{Err: "server misbehaving", IsTemporary: true}}
-		_, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeMX, 300)
-		assert.Equal(t, dns.RcodeServerFailure, rcode)
-	})
-
-	t.Run("unsupported type is NODATA", func(t *testing.T) {
-		r := &mockRecordResolver{}
-		rrs, rcode := LookupRecords(context.Background(), r, "example.com.", dns.TypeCAA, 300)
-		assert.Equal(t, dns.RcodeSuccess, rcode)
-		assert.Empty(t, rrs)
-	})
-}
-
-func TestStripOPT(t *testing.T) {
-	rm := &dns.Msg{
-		Extra: []dns.RR{
-			&dns.OPT{Hdr: dns.RR_Header{Name: ".", Rrtype: dns.TypeOPT}},
-			&dns.A{Hdr: dns.RR_Header{Name: "x.", Rrtype: dns.TypeA}, A: net.IPv4(1, 2, 3, 4)},
-		},
-	}
-	StripOPT(rm)
-	assert.Len(t, rm.Extra, 1, "OPT should be removed, A kept")
-	_, isOPT := rm.Extra[0].(*dns.OPT)
-	assert.False(t, isOPT, "remaining record must not be OPT")
-}
-
-func TestExtractEDE(t *testing.T) {
-	t.Run("no edns", func(t *testing.T) {
-		_, ok := ExtractEDE(&dns.Msg{})
-		assert.False(t, ok, "message without OPT has no EDE")
-	})
-
-	t.Run("edns without ede", func(t *testing.T) {
-		rm := &dns.Msg{}
-		rm.SetEdns0(4096, false)
-		_, ok := ExtractEDE(rm)
-		assert.False(t, ok, "OPT without EDE option returns false")
-	})
-
-	t.Run("with ede", func(t *testing.T) {
-		rm := &dns.Msg{}
-		opt := &dns.OPT{Hdr: dns.RR_Header{Name: ".", Rrtype: dns.TypeOPT}}
-		opt.Option = append(opt.Option, &dns.EDNS0_EDE{InfoCode: 49152, ExtraText: "upstream timeout"})
-		rm.Extra = append(rm.Extra, opt)
-
-		ede, ok := ExtractEDE(rm)
-		assert.True(t, ok, "EDE option should be found")
-		assert.Equal(t, uint16(49152), ede.InfoCode)
-		assert.Equal(t, "upstream timeout", ede.ExtraText)
-	})
-}

client/internal/dns/server.go

@@ -6,7 +6,6 @@ import (
 	"fmt"
 	"net/netip"
 	"net/url"
-	"os"
 	"slices"
 	"strings"
 	"sync"
@@ -39,15 +38,11 @@ const (
 	// defaultWarningDelayBase is the starting grace window before a
 	// "Nameserver group unreachable" event fires for a group that's
 	// never been healthy and only has overlay upstreams with no
-	// Connected peer. Per-server and overridable via envWarningDelay;
-	// see warningDelay.
-	defaultWarningDelayBase = 60 * time.Second
+	// Connected peer. Per-server and overridable; see warningDelayFor.
+	defaultWarningDelayBase = 30 * time.Second
 	// warningDelayBonusCap caps the route-count bonus added to the
-	// base grace window. See warningDelay.
+	// base grace window. See warningDelayFor.
 	warningDelayBonusCap = 30 * time.Second
-	// envWarningDelay overrides defaultWarningDelayBase with a Go duration
-	// string (e.g. "90s", "2m"). Invalid or non-positive values are ignored.
-	envWarningDelay = "NB_DNS_HEALTH_WARNING_DELAY"
 )
 
 // errNoUsableNameservers signals that a merged-domain group has no usable
@@ -140,7 +135,7 @@ type DefaultServer struct {
 	disableSys         bool
 	mux                sync.Mutex
 	service            service
-	dnsMuxHandlers     []handlerWrapper
+	dnsMuxMap          registeredHandlerMap
 	localResolver      *local.Resolver
 	wgInterface        WGIface
 	hostManager        hostManager
@@ -204,6 +199,8 @@ type handlerWrapper struct {
 	priority int
 }
 
+type registeredHandlerMap map[types.HandlerID]handlerWrapper
+
 // DefaultServerConfig holds configuration parameters for NewDefaultServer
 type DefaultServerConfig struct {
 	WgInterface    WGIface
@@ -292,6 +289,7 @@ func newDefaultServer(
 		service:           dnsService,
 		handlerChain:      handlerChain,
 		extraDomains:      make(map[domain.Domain]int),
+		dnsMuxMap:         make(registeredHandlerMap),
 		localResolver:     local.NewResolver(),
 		wgInterface:       wgInterface,
 		statusRecorder:    statusRecorder,
@@ -300,7 +298,7 @@ func newDefaultServer(
 		hostManager:       &noopHostConfigurator{},
 		mgmtCacheResolver: mgmtCacheResolver,
 		currentConfigHash: ^uint64(0), // Initialize to max uint64 to ensure first config is always applied
-		warningDelayBase:  warningDelayBaseFromEnv(),
+		warningDelayBase:  defaultWarningDelayBase,
 		healthRefresh:     make(chan struct{}, 1),
 	}
 	// Wire the local resolver against the peer status recorder so it can
@@ -330,7 +328,7 @@ func (s *DefaultServer) SetRouteSources(selected, active func() route.HAMap) {
 	type routeSettable interface {
 		setSelectedRoutes(func() route.HAMap)
 	}
-	for _, entry := range s.dnsMuxHandlers {
+	for _, entry := range s.dnsMuxMap {
 		if h, ok := entry.handler.(routeSettable); ok {
 			h.setSelectedRoutes(selected)
 		}
@@ -980,23 +978,19 @@ func (s *DefaultServer) usableNameServers(nameServers []nbdns.NameServer) []neti
 
 func (s *DefaultServer) updateMux(muxUpdates []handlerWrapper) {
 	// this will introduce a short period of time when the server is not able to handle DNS requests
-	for _, existing := range s.dnsMuxHandlers {
+	for _, existing := range s.dnsMuxMap {
 		s.deregisterHandler([]string{existing.domain}, existing.priority)
-		// The local resolver is a persistent singleton shared by every custom
-		// zone and reused across config updates. Its chain registrations are
-		// per-config and must be deregistered, but Stop() cancels its lookup
-		// context (breaking external CNAME-target resolution) and clears its
-		// records, so it must not be torn down here.
-		if existing.handler != s.localResolver {
-			existing.handler.Stop()
-		}
+		existing.handler.Stop()
 	}
 
+	muxUpdateMap := make(registeredHandlerMap)
+
 	for _, update := range muxUpdates {
 		s.registerHandler([]string{update.domain}, update.handler, update.priority)
+		muxUpdateMap[update.handler.ID()] = update
 	}
 
-	s.dnsMuxHandlers = muxUpdates
+	s.dnsMuxMap = muxUpdateMap
 }
 
 // updateNSGroupStates records the new group set and pokes the refresher.
@@ -1160,26 +1154,6 @@ func (s *DefaultServer) projectUnhealthy(p *nsGroupProj, servers []netip.AddrPor
 	return false
 }
 
-// warningDelayBaseFromEnv returns the base grace window, honoring
-// envWarningDelay when it holds a valid positive Go duration. Invalid or
-// non-positive values fall back to defaultWarningDelayBase.
-func warningDelayBaseFromEnv() time.Duration {
-	val := os.Getenv(envWarningDelay)
-	if val == "" {
-		return defaultWarningDelayBase
-	}
-	d, err := time.ParseDuration(val)
-	if err != nil {
-		log.Warnf("invalid %s value %q, using default %v: %v", envWarningDelay, val, defaultWarningDelayBase, err)
-		return defaultWarningDelayBase
-	}
-	if d <= 0 {
-		log.Warnf("%s must be positive, got %v, using default %v", envWarningDelay, d, defaultWarningDelayBase)
-		return defaultWarningDelayBase
-	}
-	return d
-}
-
 // warningDelay returns the grace window for the given selected-route
 // count. Scales gently: +1s per 100 routes, capped by
 // warningDelayBonusCap. Parallel handshakes mean handshake time grows
@@ -1230,7 +1204,7 @@ func (s *DefaultServer) groupHasImmediateUpstream(servers []netip.AddrPort, snap
 // in more than one handler.
 func (s *DefaultServer) collectUpstreamHealth() map[netip.AddrPort]UpstreamHealth {
 	merged := make(map[netip.AddrPort]UpstreamHealth)
-	for _, entry := range s.dnsMuxHandlers {
+	for _, entry := range s.dnsMuxMap {
 		reporter, ok := entry.handler.(upstreamHealthReporter)
 		if !ok {
 			continue

client/internal/dns/server_privileged_test.go

@@ -1,485 +0,0 @@
-//go:build privileged
-
-package dns
-
-import (
-	"context"
-	"fmt"
-	"net/netip"
-	"os"
-	"testing"
-
-	"github.com/golang/mock/gomock"
-	"github.com/miekg/dns"
-	"github.com/stretchr/testify/assert"
-	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
-
-	"github.com/netbirdio/netbird/client/iface"
-	pfmock "github.com/netbirdio/netbird/client/iface/mocks"
-	"github.com/netbirdio/netbird/client/iface/wgaddr"
-	"github.com/netbirdio/netbird/client/internal/dns/local"
-	"github.com/netbirdio/netbird/client/internal/dns/test"
-	"github.com/netbirdio/netbird/client/internal/peer"
-	"github.com/netbirdio/netbird/client/internal/stdnet"
-	nbdns "github.com/netbirdio/netbird/dns"
-)
-
-func TestUpdateDNSServer(t *testing.T) {
-
-	nameServers := []nbdns.NameServer{
-		{
-			IP:     netip.MustParseAddr("8.8.8.8"),
-			NSType: nbdns.UDPNameServerType,
-			Port:   53,
-		},
-		{
-			IP:     netip.MustParseAddr("8.8.4.4"),
-			NSType: nbdns.UDPNameServerType,
-			Port:   53,
-		},
-	}
-
-	testCases := []struct {
-		name                string
-		initUpstreamMap     []handlerWrapper
-		initLocalZones      []nbdns.CustomZone
-		initSerial          uint64
-		inputSerial         uint64
-		inputUpdate         nbdns.Config
-		shouldFail          bool
-		expectedUpstreamMap []handlerWrapper
-		expectedLocalQs     []dns.Question
-	}{
-		{
-			name:            "Initial Config Should Succeed",
-			initUpstreamMap: nil,
-			initSerial:      0,
-			inputSerial:     1,
-			inputUpdate: nbdns.Config{
-				ServiceEnable: true,
-				CustomZones: []nbdns.CustomZone{
-					{
-						Domain:  "netbird.cloud",
-						Records: zoneRecords,
-					},
-				},
-				NameServerGroups: []*nbdns.NameServerGroup{
-					{
-						Domains:     []string{"netbird.io"},
-						NameServers: nameServers,
-					},
-					{
-						NameServers: nameServers,
-						Primary:     true,
-					},
-				},
-			},
-			expectedUpstreamMap: []handlerWrapper{
-				{
-					domain:   "netbird.io",
-					priority: PriorityUpstream,
-				},
-				{
-					domain:   "netbird.cloud",
-					priority: PriorityLocal,
-				},
-				{
-					domain:   nbdns.RootZone,
-					priority: PriorityDefault,
-				},
-			},
-			expectedLocalQs: []dns.Question{{Name: "peera.netbird.cloud.", Qtype: dns.TypeA, Qclass: dns.ClassINET}},
-		},
-		{
-			name:           "New Config Should Succeed",
-			initLocalZones: []nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}},
-			initUpstreamMap: []handlerWrapper{
-				{
-					domain:   "netbird.cloud",
-					handler:  &mockHandler{},
-					priority: PriorityUpstream,
-				},
-			},
-			initSerial:  0,
-			inputSerial: 1,
-			inputUpdate: nbdns.Config{
-				ServiceEnable: true,
-				CustomZones: []nbdns.CustomZone{
-					{
-						Domain:  "netbird.cloud",
-						Records: zoneRecords,
-					},
-				},
-				NameServerGroups: []*nbdns.NameServerGroup{
-					{
-						Domains:     []string{"netbird.io"},
-						NameServers: nameServers,
-					},
-				},
-			},
-			expectedUpstreamMap: []handlerWrapper{
-				{
-					domain:   "netbird.io",
-					priority: PriorityUpstream,
-				},
-				{
-					domain:   "netbird.cloud",
-					priority: PriorityLocal,
-				},
-			},
-			expectedLocalQs: []dns.Question{{Name: zoneRecords[0].Name, Qtype: 1, Qclass: 1}},
-		},
-		{
-			name:            "Smaller Config Serial Should Be Skipped",
-			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: nil,
-			initSerial:      2,
-			inputSerial:     1,
-			shouldFail:      true,
-		},
-		{
-			name:            "Empty NS Group Domain Or Not Primary Element Should Fail",
-			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: nil,
-			initSerial:      0,
-			inputSerial:     1,
-			inputUpdate: nbdns.Config{
-				ServiceEnable: true,
-				CustomZones: []nbdns.CustomZone{
-					{
-						Domain:  "netbird.cloud",
-						Records: zoneRecords,
-					},
-				},
-				NameServerGroups: []*nbdns.NameServerGroup{
-					{
-						NameServers: nameServers,
-					},
-				},
-			},
-			shouldFail: true,
-		},
-		{
-			name:            "Invalid NS Group Nameservers list Should Fail",
-			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: nil,
-			initSerial:      0,
-			inputSerial:     1,
-			inputUpdate: nbdns.Config{
-				ServiceEnable: true,
-				CustomZones: []nbdns.CustomZone{
-					{
-						Domain:  "netbird.cloud",
-						Records: zoneRecords,
-					},
-				},
-				NameServerGroups: []*nbdns.NameServerGroup{
-					{
-						NameServers: nameServers,
-					},
-				},
-			},
-			shouldFail: true,
-		},
-		{
-			name:            "Invalid Custom Zone Records list Should Skip",
-			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: nil,
-			initSerial:      0,
-			inputSerial:     1,
-			inputUpdate: nbdns.Config{
-				ServiceEnable: true,
-				CustomZones: []nbdns.CustomZone{
-					{
-						Domain: "netbird.cloud",
-					},
-				},
-				NameServerGroups: []*nbdns.NameServerGroup{
-					{
-						NameServers: nameServers,
-						Primary:     true,
-					},
-				},
-			},
-			expectedUpstreamMap: []handlerWrapper{{
-				domain:   ".",
-				priority: PriorityDefault,
-			}},
-		},
-		{
-			name:           "Empty Config Should Succeed and Clean Maps",
-			initLocalZones: []nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: int(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}},
-			initUpstreamMap: []handlerWrapper{
-				{
-					domain:   zoneRecords[0].Name,
-					handler:  &mockHandler{},
-					priority: PriorityUpstream,
-				},
-			},
-			initSerial:          0,
-			inputSerial:         1,
-			inputUpdate:         nbdns.Config{ServiceEnable: true},
-			expectedUpstreamMap: nil,
-			expectedLocalQs:     []dns.Question{},
-		},
-		{
-			name:           "Disabled Service Should clean map",
-			initLocalZones: []nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: int(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}},
-			initUpstreamMap: []handlerWrapper{
-				{
-					domain:   zoneRecords[0].Name,
-					handler:  &mockHandler{},
-					priority: PriorityUpstream,
-				},
-			},
-			initSerial:          0,
-			inputSerial:         1,
-			inputUpdate:         nbdns.Config{ServiceEnable: false},
-			expectedUpstreamMap: nil,
-			expectedLocalQs:     []dns.Question{},
-		},
-	}
-
-	for n, testCase := range testCases {
-		t.Run(testCase.name, func(t *testing.T) {
-			privKey, _ := wgtypes.GenerateKey()
-			newNet, err := stdnet.NewNet(context.Background(), nil)
-			if err != nil {
-				t.Fatal(err)
-			}
-
-			opts := iface.WGIFaceOpts{
-				IFaceName:    fmt.Sprintf("utun230%d", n),
-				Address:      wgaddr.MustParseWGAddress(fmt.Sprintf("100.66.100.%d/32", n+1)),
-				WGPort:       33100,
-				WGPrivKey:    privKey.String(),
-				MTU:          iface.DefaultMTU,
-				TransportNet: newNet,
-			}
-
-			wgIface, err := iface.NewWGIFace(opts)
-			if err != nil {
-				t.Fatal(err)
-			}
-			err = wgIface.Create()
-			if err != nil {
-				t.Fatal(err)
-			}
-			defer func() {
-				err = wgIface.Close()
-				if err != nil {
-					t.Log(err)
-				}
-			}()
-			dnsServer, err := NewDefaultServer(context.Background(), DefaultServerConfig{
-				WgInterface:    wgIface,
-				CustomAddress:  "",
-				StatusRecorder: peer.NewRecorder("mgm"),
-				StateManager:   nil,
-				DisableSys:     false,
-			})
-			if err != nil {
-				t.Fatal(err)
-			}
-			err = dnsServer.Initialize()
-			if err != nil {
-				t.Fatal(err)
-			}
-			defer func() {
-				err = dnsServer.hostManager.restoreHostDNS()
-				if err != nil {
-					t.Log(err)
-				}
-			}()
-
-			dnsServer.dnsMuxHandlers = testCase.initUpstreamMap
-			dnsServer.localResolver.Update(testCase.initLocalZones)
-			dnsServer.updateSerial = testCase.initSerial
-
-			err = dnsServer.UpdateDNSServer(testCase.inputSerial, testCase.inputUpdate)
-			if err != nil {
-				if testCase.shouldFail {
-					return
-				}
-				t.Fatalf("update dns server should not fail, got error: %v", err)
-			}
-
-			if len(dnsServer.dnsMuxHandlers) != len(testCase.expectedUpstreamMap) {
-				t.Fatalf("update upstream failed, map size is different than expected, want %d, got %d", len(testCase.expectedUpstreamMap), len(dnsServer.dnsMuxHandlers))
-			}
-
-			for _, expected := range testCase.expectedUpstreamMap {
-				found := false
-				for _, got := range dnsServer.dnsMuxHandlers {
-					if got.domain == expected.domain && got.priority == expected.priority {
-						found = true
-						break
-					}
-				}
-				if !found {
-					t.Fatalf("update upstream failed, handler for domain=%s priority=%d not found in dnsMuxHandlers: %#v", expected.domain, expected.priority, dnsServer.dnsMuxHandlers)
-				}
-			}
-
-			var responseMSG *dns.Msg
-			responseWriter := &test.MockResponseWriter{
-				WriteMsgFunc: func(m *dns.Msg) error {
-					responseMSG = m
-					return nil
-				},
-			}
-			for _, q := range testCase.expectedLocalQs {
-				dnsServer.localResolver.ServeDNS(responseWriter, &dns.Msg{
-					Question: []dns.Question{q},
-				})
-			}
-
-			if len(testCase.expectedLocalQs) > 0 {
-				assert.NotNil(t, responseMSG, "response message should not be nil")
-				assert.Equal(t, dns.RcodeSuccess, responseMSG.Rcode, "response code should be success")
-				assert.NotEmpty(t, responseMSG.Answer, "response message should have answers")
-			}
-		})
-	}
-}
-
-func TestDNSFakeResolverHandleUpdates(t *testing.T) {
-	ov := os.Getenv("NB_WG_KERNEL_DISABLED")
-	defer t.Setenv("NB_WG_KERNEL_DISABLED", ov)
-
-	t.Setenv("NB_WG_KERNEL_DISABLED", "true")
-	newNet, err := stdnet.NewNet(context.Background(), []string{"utun2301"})
-	if err != nil {
-		t.Errorf("create stdnet: %v", err)
-		return
-	}
-
-	privKey, _ := wgtypes.GeneratePrivateKey()
-	opts := iface.WGIFaceOpts{
-		IFaceName:    "utun2301",
-		Address:      wgaddr.MustParseWGAddress("100.66.100.1/32"),
-		WGPort:       33100,
-		WGPrivKey:    privKey.String(),
-		MTU:          iface.DefaultMTU,
-		TransportNet: newNet,
-	}
-	wgIface, err := iface.NewWGIFace(opts)
-	if err != nil {
-		t.Errorf("build interface wireguard: %v", err)
-		return
-	}
-
-	err = wgIface.Create()
-	if err != nil {
-		t.Errorf("create and init wireguard interface: %v", err)
-		return
-	}
-	defer func() {
-		if err = wgIface.Close(); err != nil {
-			t.Logf("close wireguard interface: %v", err)
-		}
-	}()
-
-	ctrl := gomock.NewController(t)
-	defer ctrl.Finish()
-
-	packetfilter := pfmock.NewMockPacketFilter(ctrl)
-	packetfilter.EXPECT().FilterOutbound(gomock.Any(), gomock.Any()).AnyTimes()
-	packetfilter.EXPECT().SetUDPPacketHook(gomock.Any(), gomock.Any(), gomock.Any()).AnyTimes()
-	packetfilter.EXPECT().SetTCPPacketHook(gomock.Any(), gomock.Any(), gomock.Any()).AnyTimes()
-
-	if err := wgIface.SetFilter(packetfilter); err != nil {
-		t.Errorf("set packet filter: %v", err)
-		return
-	}
-
-	dnsServer, err := NewDefaultServer(context.Background(), DefaultServerConfig{
-		WgInterface:    wgIface,
-		CustomAddress:  "",
-		StatusRecorder: peer.NewRecorder("mgm"),
-		StateManager:   nil,
-		DisableSys:     false,
-	})
-	if err != nil {
-		t.Errorf("create DNS server: %v", err)
-		return
-	}
-
-	err = dnsServer.Initialize()
-	if err != nil {
-		t.Errorf("run DNS server: %v", err)
-		return
-	}
-	defer func() {
-		if err = dnsServer.hostManager.restoreHostDNS(); err != nil {
-			t.Logf("restore DNS settings on the host: %v", err)
-			return
-		}
-	}()
-
-	dnsServer.dnsMuxHandlers = []handlerWrapper{
-		{
-			domain:   zoneRecords[0].Name,
-			handler:  &local.Resolver{},
-			priority: PriorityUpstream,
-		},
-	}
-	dnsServer.localResolver.Update([]nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: int(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}})
-	dnsServer.updateSerial = 0
-
-	nameServers := []nbdns.NameServer{
-		{
-			IP:     netip.MustParseAddr("8.8.8.8"),
-			NSType: nbdns.UDPNameServerType,
-			Port:   53,
-		},
-		{
-			IP:     netip.MustParseAddr("8.8.4.4"),
-			NSType: nbdns.UDPNameServerType,
-			Port:   53,
-		},
-	}
-
-	update := nbdns.Config{
-		ServiceEnable: true,
-		CustomZones: []nbdns.CustomZone{
-			{
-				Domain:  "netbird.cloud",
-				Records: zoneRecords,
-			},
-		},
-		NameServerGroups: []*nbdns.NameServerGroup{
-			{
-				Domains:     []string{"netbird.io"},
-				NameServers: nameServers,
-			},
-			{
-				NameServers: nameServers,
-				Primary:     true,
-			},
-		},
-	}
-
-	// Start the server with regular configuration
-	if err := dnsServer.UpdateDNSServer(1, update); err != nil {
-		t.Fatalf("update dns server should not fail, got error: %v", err)
-		return
-	}
-
-	update2 := update
-	update2.ServiceEnable = false
-	// Disable the server, stop the listener
-	if err := dnsServer.UpdateDNSServer(2, update2); err != nil {
-		t.Fatalf("update dns server should not fail, got error: %v", err)
-		return
-	}
-
-	update3 := update2
-	update3.NameServerGroups = update3.NameServerGroups[:1]
-	// But service still get updates and we checking that we handle
-	// internal state in the right way
-	if err := dnsServer.UpdateDNSServer(3, update3); err != nil {
-		t.Fatalf("update dns server should not fail, got error: %v", err)
-		return
-	}
-}

client/internal/dns/server_test.go

@@ -10,6 +10,7 @@ import (
 	"testing"
 	"time"
 
+	"github.com/golang/mock/gomock"
 	"github.com/miekg/dns"
 	log "github.com/sirupsen/logrus"
 	"github.com/stretchr/testify/assert"
@@ -22,6 +23,7 @@ import (
 	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/iface/configurer"
 	"github.com/netbirdio/netbird/client/iface/device"
+	pfmock "github.com/netbirdio/netbird/client/iface/mocks"
 	"github.com/netbirdio/netbird/client/iface/wgaddr"
 	"github.com/netbirdio/netbird/client/internal/dns/local"
 	"github.com/netbirdio/netbird/client/internal/dns/test"
@@ -102,6 +104,481 @@ func init() {
 	formatter.SetTextFormatter(log.StandardLogger())
 }
 
+func generateDummyHandler(d string, servers []nbdns.NameServer) *upstreamResolverBase {
+	var srvs []netip.AddrPort
+	for _, srv := range servers {
+		srvs = append(srvs, srv.AddrPort())
+	}
+	u := &upstreamResolverBase{
+		domain: domain.Domain(d),
+		cancel: func() {},
+	}
+	u.addRace(srvs)
+	return u
+}
+
+func TestUpdateDNSServer(t *testing.T) {
+
+	nameServers := []nbdns.NameServer{
+		{
+			IP:     netip.MustParseAddr("8.8.8.8"),
+			NSType: nbdns.UDPNameServerType,
+			Port:   53,
+		},
+		{
+			IP:     netip.MustParseAddr("8.8.4.4"),
+			NSType: nbdns.UDPNameServerType,
+			Port:   53,
+		},
+	}
+
+	dummyHandler := local.NewResolver()
+
+	testCases := []struct {
+		name                string
+		initUpstreamMap     registeredHandlerMap
+		initLocalZones      []nbdns.CustomZone
+		initSerial          uint64
+		inputSerial         uint64
+		inputUpdate         nbdns.Config
+		shouldFail          bool
+		expectedUpstreamMap registeredHandlerMap
+		expectedLocalQs     []dns.Question
+	}{
+		{
+			name:            "Initial Config Should Succeed",
+			initUpstreamMap: make(registeredHandlerMap),
+			initSerial:      0,
+			inputSerial:     1,
+			inputUpdate: nbdns.Config{
+				ServiceEnable: true,
+				CustomZones: []nbdns.CustomZone{
+					{
+						Domain:  "netbird.cloud",
+						Records: zoneRecords,
+					},
+				},
+				NameServerGroups: []*nbdns.NameServerGroup{
+					{
+						Domains:     []string{"netbird.io"},
+						NameServers: nameServers,
+					},
+					{
+						NameServers: nameServers,
+						Primary:     true,
+					},
+				},
+			},
+			expectedUpstreamMap: registeredHandlerMap{
+				generateDummyHandler("netbird.io", nameServers).ID(): handlerWrapper{
+					domain:   "netbird.io",
+					handler:  dummyHandler,
+					priority: PriorityUpstream,
+				},
+				dummyHandler.ID(): handlerWrapper{
+					domain:   "netbird.cloud",
+					handler:  dummyHandler,
+					priority: PriorityLocal,
+				},
+				generateDummyHandler(".", nameServers).ID(): handlerWrapper{
+					domain:   nbdns.RootZone,
+					handler:  dummyHandler,
+					priority: PriorityDefault,
+				},
+			},
+			expectedLocalQs: []dns.Question{{Name: "peera.netbird.cloud.", Qtype: dns.TypeA, Qclass: dns.ClassINET}},
+		},
+		{
+			name:           "New Config Should Succeed",
+			initLocalZones: []nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}},
+			initUpstreamMap: registeredHandlerMap{
+				generateDummyHandler(zoneRecords[0].Name, nameServers).ID(): handlerWrapper{
+					domain:   "netbird.cloud",
+					handler:  dummyHandler,
+					priority: PriorityUpstream,
+				},
+			},
+			initSerial:  0,
+			inputSerial: 1,
+			inputUpdate: nbdns.Config{
+				ServiceEnable: true,
+				CustomZones: []nbdns.CustomZone{
+					{
+						Domain:  "netbird.cloud",
+						Records: zoneRecords,
+					},
+				},
+				NameServerGroups: []*nbdns.NameServerGroup{
+					{
+						Domains:     []string{"netbird.io"},
+						NameServers: nameServers,
+					},
+				},
+			},
+			expectedUpstreamMap: registeredHandlerMap{
+				generateDummyHandler("netbird.io", nameServers).ID(): handlerWrapper{
+					domain:   "netbird.io",
+					handler:  dummyHandler,
+					priority: PriorityUpstream,
+				},
+				"local-resolver": handlerWrapper{
+					domain:   "netbird.cloud",
+					handler:  dummyHandler,
+					priority: PriorityLocal,
+				},
+			},
+			expectedLocalQs: []dns.Question{{Name: zoneRecords[0].Name, Qtype: 1, Qclass: 1}},
+		},
+		{
+			name:            "Smaller Config Serial Should Be Skipped",
+			initLocalZones:  []nbdns.CustomZone{},
+			initUpstreamMap: make(registeredHandlerMap),
+			initSerial:      2,
+			inputSerial:     1,
+			shouldFail:      true,
+		},
+		{
+			name:            "Empty NS Group Domain Or Not Primary Element Should Fail",
+			initLocalZones:  []nbdns.CustomZone{},
+			initUpstreamMap: make(registeredHandlerMap),
+			initSerial:      0,
+			inputSerial:     1,
+			inputUpdate: nbdns.Config{
+				ServiceEnable: true,
+				CustomZones: []nbdns.CustomZone{
+					{
+						Domain:  "netbird.cloud",
+						Records: zoneRecords,
+					},
+				},
+				NameServerGroups: []*nbdns.NameServerGroup{
+					{
+						NameServers: nameServers,
+					},
+				},
+			},
+			shouldFail: true,
+		},
+		{
+			name:            "Invalid NS Group Nameservers list Should Fail",
+			initLocalZones:  []nbdns.CustomZone{},
+			initUpstreamMap: make(registeredHandlerMap),
+			initSerial:      0,
+			inputSerial:     1,
+			inputUpdate: nbdns.Config{
+				ServiceEnable: true,
+				CustomZones: []nbdns.CustomZone{
+					{
+						Domain:  "netbird.cloud",
+						Records: zoneRecords,
+					},
+				},
+				NameServerGroups: []*nbdns.NameServerGroup{
+					{
+						NameServers: nameServers,
+					},
+				},
+			},
+			shouldFail: true,
+		},
+		{
+			name:            "Invalid Custom Zone Records list Should Skip",
+			initLocalZones:  []nbdns.CustomZone{},
+			initUpstreamMap: make(registeredHandlerMap),
+			initSerial:      0,
+			inputSerial:     1,
+			inputUpdate: nbdns.Config{
+				ServiceEnable: true,
+				CustomZones: []nbdns.CustomZone{
+					{
+						Domain: "netbird.cloud",
+					},
+				},
+				NameServerGroups: []*nbdns.NameServerGroup{
+					{
+						NameServers: nameServers,
+						Primary:     true,
+					},
+				},
+			},
+			expectedUpstreamMap: registeredHandlerMap{generateDummyHandler(".", nameServers).ID(): handlerWrapper{
+				domain:   ".",
+				handler:  dummyHandler,
+				priority: PriorityDefault,
+			}},
+		},
+		{
+			name:           "Empty Config Should Succeed and Clean Maps",
+			initLocalZones: []nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: int(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}},
+			initUpstreamMap: registeredHandlerMap{
+				generateDummyHandler(zoneRecords[0].Name, nameServers).ID(): handlerWrapper{
+					domain:   zoneRecords[0].Name,
+					handler:  dummyHandler,
+					priority: PriorityUpstream,
+				},
+			},
+			initSerial:          0,
+			inputSerial:         1,
+			inputUpdate:         nbdns.Config{ServiceEnable: true},
+			expectedUpstreamMap: make(registeredHandlerMap),
+			expectedLocalQs:     []dns.Question{},
+		},
+		{
+			name:           "Disabled Service Should clean map",
+			initLocalZones: []nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: int(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}},
+			initUpstreamMap: registeredHandlerMap{
+				generateDummyHandler(zoneRecords[0].Name, nameServers).ID(): handlerWrapper{
+					domain:   zoneRecords[0].Name,
+					handler:  dummyHandler,
+					priority: PriorityUpstream,
+				},
+			},
+			initSerial:          0,
+			inputSerial:         1,
+			inputUpdate:         nbdns.Config{ServiceEnable: false},
+			expectedUpstreamMap: make(registeredHandlerMap),
+			expectedLocalQs:     []dns.Question{},
+		},
+	}
+
+	for n, testCase := range testCases {
+		t.Run(testCase.name, func(t *testing.T) {
+			privKey, _ := wgtypes.GenerateKey()
+			newNet, err := stdnet.NewNet(context.Background(), nil)
+			if err != nil {
+				t.Fatal(err)
+			}
+
+			opts := iface.WGIFaceOpts{
+				IFaceName:    fmt.Sprintf("utun230%d", n),
+				Address:      wgaddr.MustParseWGAddress(fmt.Sprintf("100.66.100.%d/32", n+1)),
+				WGPort:       33100,
+				WGPrivKey:    privKey.String(),
+				MTU:          iface.DefaultMTU,
+				TransportNet: newNet,
+			}
+
+			wgIface, err := iface.NewWGIFace(opts)
+			if err != nil {
+				t.Fatal(err)
+			}
+			err = wgIface.Create()
+			if err != nil {
+				t.Fatal(err)
+			}
+			defer func() {
+				err = wgIface.Close()
+				if err != nil {
+					t.Log(err)
+				}
+			}()
+			dnsServer, err := NewDefaultServer(context.Background(), DefaultServerConfig{
+				WgInterface:    wgIface,
+				CustomAddress:  "",
+				StatusRecorder: peer.NewRecorder("mgm"),
+				StateManager:   nil,
+				DisableSys:     false,
+			})
+			if err != nil {
+				t.Fatal(err)
+			}
+			err = dnsServer.Initialize()
+			if err != nil {
+				t.Fatal(err)
+			}
+			defer func() {
+				err = dnsServer.hostManager.restoreHostDNS()
+				if err != nil {
+					t.Log(err)
+				}
+			}()
+
+			dnsServer.dnsMuxMap = testCase.initUpstreamMap
+			dnsServer.localResolver.Update(testCase.initLocalZones)
+			dnsServer.updateSerial = testCase.initSerial
+
+			err = dnsServer.UpdateDNSServer(testCase.inputSerial, testCase.inputUpdate)
+			if err != nil {
+				if testCase.shouldFail {
+					return
+				}
+				t.Fatalf("update dns server should not fail, got error: %v", err)
+			}
+
+			if len(dnsServer.dnsMuxMap) != len(testCase.expectedUpstreamMap) {
+				t.Fatalf("update upstream failed, map size is different than expected, want %d, got %d", len(testCase.expectedUpstreamMap), len(dnsServer.dnsMuxMap))
+			}
+
+			for key := range testCase.expectedUpstreamMap {
+				_, found := dnsServer.dnsMuxMap[key]
+				if !found {
+					t.Fatalf("update upstream failed, key %s was not found in the dnsMuxMap: %#v", key, dnsServer.dnsMuxMap)
+				}
+			}
+
+			var responseMSG *dns.Msg
+			responseWriter := &test.MockResponseWriter{
+				WriteMsgFunc: func(m *dns.Msg) error {
+					responseMSG = m
+					return nil
+				},
+			}
+			for _, q := range testCase.expectedLocalQs {
+				dnsServer.localResolver.ServeDNS(responseWriter, &dns.Msg{
+					Question: []dns.Question{q},
+				})
+			}
+
+			if len(testCase.expectedLocalQs) > 0 {
+				assert.NotNil(t, responseMSG, "response message should not be nil")
+				assert.Equal(t, dns.RcodeSuccess, responseMSG.Rcode, "response code should be success")
+				assert.NotEmpty(t, responseMSG.Answer, "response message should have answers")
+			}
+		})
+	}
+}
+
+func TestDNSFakeResolverHandleUpdates(t *testing.T) {
+	ov := os.Getenv("NB_WG_KERNEL_DISABLED")
+	defer t.Setenv("NB_WG_KERNEL_DISABLED", ov)
+
+	t.Setenv("NB_WG_KERNEL_DISABLED", "true")
+	newNet, err := stdnet.NewNet(context.Background(), []string{"utun2301"})
+	if err != nil {
+		t.Errorf("create stdnet: %v", err)
+		return
+	}
+
+	privKey, _ := wgtypes.GeneratePrivateKey()
+	opts := iface.WGIFaceOpts{
+		IFaceName:    "utun2301",
+		Address:      wgaddr.MustParseWGAddress("100.66.100.1/32"),
+		WGPort:       33100,
+		WGPrivKey:    privKey.String(),
+		MTU:          iface.DefaultMTU,
+		TransportNet: newNet,
+	}
+	wgIface, err := iface.NewWGIFace(opts)
+	if err != nil {
+		t.Errorf("build interface wireguard: %v", err)
+		return
+	}
+
+	err = wgIface.Create()
+	if err != nil {
+		t.Errorf("create and init wireguard interface: %v", err)
+		return
+	}
+	defer func() {
+		if err = wgIface.Close(); err != nil {
+			t.Logf("close wireguard interface: %v", err)
+		}
+	}()
+
+	ctrl := gomock.NewController(t)
+	defer ctrl.Finish()
+
+	packetfilter := pfmock.NewMockPacketFilter(ctrl)
+	packetfilter.EXPECT().FilterOutbound(gomock.Any(), gomock.Any()).AnyTimes()
+	packetfilter.EXPECT().SetUDPPacketHook(gomock.Any(), gomock.Any(), gomock.Any()).AnyTimes()
+	packetfilter.EXPECT().SetTCPPacketHook(gomock.Any(), gomock.Any(), gomock.Any()).AnyTimes()
+
+	if err := wgIface.SetFilter(packetfilter); err != nil {
+		t.Errorf("set packet filter: %v", err)
+		return
+	}
+
+	dnsServer, err := NewDefaultServer(context.Background(), DefaultServerConfig{
+		WgInterface:    wgIface,
+		CustomAddress:  "",
+		StatusRecorder: peer.NewRecorder("mgm"),
+		StateManager:   nil,
+		DisableSys:     false,
+	})
+	if err != nil {
+		t.Errorf("create DNS server: %v", err)
+		return
+	}
+
+	err = dnsServer.Initialize()
+	if err != nil {
+		t.Errorf("run DNS server: %v", err)
+		return
+	}
+	defer func() {
+		if err = dnsServer.hostManager.restoreHostDNS(); err != nil {
+			t.Logf("restore DNS settings on the host: %v", err)
+			return
+		}
+	}()
+
+	dnsServer.dnsMuxMap = registeredHandlerMap{
+		"id1": handlerWrapper{
+			domain:   zoneRecords[0].Name,
+			handler:  &local.Resolver{},
+			priority: PriorityUpstream,
+		},
+	}
+	dnsServer.localResolver.Update([]nbdns.CustomZone{{Domain: "netbird.cloud", Records: []nbdns.SimpleRecord{{Name: "netbird.cloud", Type: int(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.0.0.1"}}}})
+	dnsServer.updateSerial = 0
+
+	nameServers := []nbdns.NameServer{
+		{
+			IP:     netip.MustParseAddr("8.8.8.8"),
+			NSType: nbdns.UDPNameServerType,
+			Port:   53,
+		},
+		{
+			IP:     netip.MustParseAddr("8.8.4.4"),
+			NSType: nbdns.UDPNameServerType,
+			Port:   53,
+		},
+	}
+
+	update := nbdns.Config{
+		ServiceEnable: true,
+		CustomZones: []nbdns.CustomZone{
+			{
+				Domain:  "netbird.cloud",
+				Records: zoneRecords,
+			},
+		},
+		NameServerGroups: []*nbdns.NameServerGroup{
+			{
+				Domains:     []string{"netbird.io"},
+				NameServers: nameServers,
+			},
+			{
+				NameServers: nameServers,
+				Primary:     true,
+			},
+		},
+	}
+
+	// Start the server with regular configuration
+	if err := dnsServer.UpdateDNSServer(1, update); err != nil {
+		t.Fatalf("update dns server should not fail, got error: %v", err)
+		return
+	}
+
+	update2 := update
+	update2.ServiceEnable = false
+	// Disable the server, stop the listener
+	if err := dnsServer.UpdateDNSServer(2, update2); err != nil {
+		t.Fatalf("update dns server should not fail, got error: %v", err)
+		return
+	}
+
+	update3 := update2
+	update3.NameServerGroups = update3.NameServerGroups[:1]
+	// But service still get updates and we checking that we handle
+	// internal state in the right way
+	if err := dnsServer.UpdateDNSServer(3, update3); err != nil {
+		t.Fatalf("update dns server should not fail, got error: %v", err)
+		return
+	}
+}
+
 func TestDNSServerStartStop(t *testing.T) {
 	testCases := []struct {
 		name     string
@@ -552,15 +1029,15 @@ func (m *mockService) RegisterMux(string, dns.Handler) {}
 func (m *mockService) DeregisterMux(string)            {}
 
 func TestDefaultServer_UpdateMux(t *testing.T) {
-	baseMatchHandlers := []handlerWrapper{
-		{
+	baseMatchHandlers := registeredHandlerMap{
+		"upstream-group1": {
 			domain: "example.com",
 			handler: &mockHandler{
 				Id: "upstream-group1",
 			},
 			priority: PriorityUpstream,
 		},
-		{
+		"upstream-group2": {
 			domain: "example.com",
 			handler: &mockHandler{
 				Id: "upstream-group2",
@@ -569,15 +1046,15 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 		},
 	}
 
-	baseRootHandlers := []handlerWrapper{
-		{
+	baseRootHandlers := registeredHandlerMap{
+		"upstream-root1": {
 			domain: ".",
 			handler: &mockHandler{
 				Id: "upstream-root1",
 			},
 			priority: PriorityDefault,
 		},
-		{
+		"upstream-root2": {
 			domain: ".",
 			handler: &mockHandler{
 				Id: "upstream-root2",
@@ -586,22 +1063,22 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 		},
 	}
 
-	baseMixedHandlers := []handlerWrapper{
-		{
+	baseMixedHandlers := registeredHandlerMap{
+		"upstream-group1": {
 			domain: "example.com",
 			handler: &mockHandler{
 				Id: "upstream-group1",
 			},
 			priority: PriorityUpstream,
 		},
-		{
+		"upstream-group2": {
 			domain: "example.com",
 			handler: &mockHandler{
 				Id: "upstream-group2",
 			},
 			priority: PriorityUpstream - 1,
 		},
-		{
+		"upstream-other": {
 			domain: "other.com",
 			handler: &mockHandler{
 				Id: "upstream-other",
@@ -612,7 +1089,7 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 
 	tests := []struct {
 		name             string
-		initialHandlers  []handlerWrapper
+		initialHandlers  registeredHandlerMap
 		updates          []handlerWrapper
 		expectedHandlers map[string]string // map[HandlerID]domain
 		description      string
@@ -896,38 +1373,32 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			server := &DefaultServer{
-				dnsMuxHandlers: tt.initialHandlers,
-				handlerChain:   NewHandlerChain(),
-				service:        &mockService{},
+				dnsMuxMap:    tt.initialHandlers,
+				handlerChain: NewHandlerChain(),
+				service:      &mockService{},
 			}
 
 			// Perform the update
 			server.updateMux(tt.updates)
 
 			// Verify the results
-			assert.Equal(t, len(tt.expectedHandlers), len(server.dnsMuxHandlers),
+			assert.Equal(t, len(tt.expectedHandlers), len(server.dnsMuxMap),
 				"Number of handlers after update doesn't match expected")
 
 			// Check each expected handler
 			for id, expectedDomain := range tt.expectedHandlers {
-				var found *handlerWrapper
-				for i := range server.dnsMuxHandlers {
-					if server.dnsMuxHandlers[i].handler.ID() == types.HandlerID(id) {
-						found = &server.dnsMuxHandlers[i]
-						break
-					}
-				}
-				assert.NotNil(t, found, "Expected handler %s not found", id)
-				if found != nil {
-					assert.Equal(t, expectedDomain, found.domain,
+				handler, exists := server.dnsMuxMap[types.HandlerID(id)]
+				assert.True(t, exists, "Expected handler %s not found", id)
+				if exists {
+					assert.Equal(t, expectedDomain, handler.domain,
 						"Domain mismatch for handler %s", id)
 				}
 			}
 
 			// Verify no unexpected handlers exist
-			for _, entry := range server.dnsMuxHandlers {
-				_, expected := tt.expectedHandlers[string(entry.handler.ID())]
-				assert.True(t, expected, "Unexpected handler found: %s", entry.handler.ID())
+			for HandlerID := range server.dnsMuxMap {
+				_, expected := tt.expectedHandlers[string(HandlerID)]
+				assert.True(t, expected, "Unexpected handler found: %s", HandlerID)
 			}
 
 			// Verify the handlerChain state and order
@@ -942,7 +1413,7 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 
 				// Verify handler exists in mux
 				foundInMux := false
-				for _, muxEntry := range server.dnsMuxHandlers {
+				for _, muxEntry := range server.dnsMuxMap {
 					if chainEntry.Handler == muxEntry.handler &&
 						chainEntry.Priority == muxEntry.priority &&
 						chainEntry.Pattern == dns.Fqdn(muxEntry.domain) {
@@ -951,108 +1422,12 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 					}
 				}
 				assert.True(t, foundInMux,
-					"Handler in chain not found in dnsMuxHandlers")
+					"Handler in chain not found in dnsMuxMap")
 			}
 		})
 	}
 }
 
-// chainHasPattern reports whether the handler chain holds an entry registered
-// for the given fqdn pattern at the given priority.
-func chainHasPattern(s *DefaultServer, pattern string, priority int) bool {
-	for _, h := range s.handlerChain.handlers {
-		if h.OrigPattern == pattern && h.Priority == priority {
-			return true
-		}
-	}
-	return false
-}
-
-// TestDefaultServer_UpdateMux_SharedHandlerZoneRemoval verifies that updateMux
-// tracks each (handler, domain) registration independently when one handler
-// serves multiple zones. Every custom zone is served by the same handler
-// instance (the local resolver, whose ID is the constant "local-resolver"), so
-// removing one zone must deregister exactly that zone's chain entry and leave
-// the others in place. Tracking registrations by handler ID alone collapses all
-// zones onto one entry, leaving removed zones in the chain to answer
-// authoritatively with no records.
-func TestDefaultServer_UpdateMux_SharedHandlerZoneRemoval(t *testing.T) {
-	// One handler serves every custom zone, mirroring s.localResolver.
-	shared := &mockHandler{Id: "local-resolver"}
-
-	server := &DefaultServer{
-		handlerChain: NewHandlerChain(),
-		service:      &mockService{},
-	}
-
-	// Two custom zones under the same handler. The surviving zone is registered
-	// last, mirroring the management emission order.
-	server.updateMux([]handlerWrapper{
-		{domain: "userzone.test", handler: shared, priority: PriorityLocal},
-		{domain: "peerzone.test", handler: shared, priority: PriorityLocal},
-	})
-
-	require.True(t, chainHasPattern(server, "userzone.test.", PriorityLocal),
-		"userzone.test should be registered after the first update")
-	require.True(t, chainHasPattern(server, "peerzone.test.", PriorityLocal),
-		"peerzone.test should be registered after the first update")
-
-	// Remove one zone, keep the other.
-	server.updateMux([]handlerWrapper{
-		{domain: "peerzone.test", handler: shared, priority: PriorityLocal},
-	})
-
-	assert.True(t, chainHasPattern(server, "peerzone.test.", PriorityLocal),
-		"peerzone.test should remain after removing userzone.test")
-	assert.False(t, chainHasPattern(server, "userzone.test.", PriorityLocal),
-		"userzone.test handler must be deregistered, not leaked in the chain")
-}
-
-// TestDefaultServer_UpdateMux_PreservesLocalResolver verifies that updateMux
-// does not tear down the shared local resolver during reconfiguration. The
-// resolver is a process-lifetime singleton reused across config updates;
-// Stop() cancels its lookup context (breaking external CNAME-target
-// resolution) and clears its records. updateMux must deregister its chain
-// entries without stopping it. Records surviving a teardown update is the
-// observable proxy: Stop() would have cleared them.
-func TestDefaultServer_UpdateMux_PreservesLocalResolver(t *testing.T) {
-	resolver := local.NewResolver()
-	require.NoError(t, resolver.RegisterRecord(nbdns.SimpleRecord{
-		Name:  "peer.netbird.cloud.",
-		Type:  int(dns.TypeA),
-		Class: nbdns.DefaultClass,
-		TTL:   300,
-		RData: "10.0.0.1",
-	}))
-
-	server := &DefaultServer{
-		handlerChain:  NewHandlerChain(),
-		service:       &mockService{},
-		localResolver: resolver,
-	}
-
-	server.updateMux([]handlerWrapper{
-		{domain: "netbird.cloud", handler: resolver, priority: PriorityLocal},
-	})
-
-	// Remove the zone. The resolver must survive so its records and lookup
-	// context stay intact for the next registration.
-	server.updateMux(nil)
-
-	var response *dns.Msg
-	resolver.ServeDNS(&test.MockResponseWriter{
-		WriteMsgFunc: func(m *dns.Msg) error {
-			response = m
-			return nil
-		},
-	}, &dns.Msg{Question: []dns.Question{{Name: "peer.netbird.cloud.", Qtype: dns.TypeA, Qclass: dns.ClassINET}}})
-
-	require.NotNil(t, response, "local resolver should answer after teardown")
-	assert.Equal(t, dns.RcodeSuccess, response.Rcode,
-		"local resolver records must survive teardown; updateMux must not Stop() the shared resolver")
-	assert.NotEmpty(t, response.Answer, "answer should contain the surviving record")
-}
-
 func TestExtraDomains(t *testing.T) {
 	tests := []struct {
 		name                string
@@ -1674,6 +2049,7 @@ func TestBuildUpstreamHandler_MergesGroupsPerDomain(t *testing.T) {
 		localResolver: local.NewResolver(),
 		handlerChain:  NewHandlerChain(),
 		hostManager:   &noopHostConfigurator{},
+		dnsMuxMap:     make(registeredHandlerMap),
 	}
 
 	groups := []*nbdns.NameServerGroup{
@@ -1831,7 +2207,7 @@ func TestEvaluateNSGroupHealth(t *testing.T) {
 	}
 }
 
-// healthStubHandler is a minimal dnsMuxHandlers entry that exposes a fixed
+// healthStubHandler is a minimal dnsMuxMap entry that exposes a fixed
 // UpstreamHealth snapshot, letting tests drive recomputeNSGroupStates
 // without spinning up real handlers.
 type healthStubHandler struct {
@@ -1907,11 +2283,12 @@ func newProjTestFixture(t *testing.T) *projTestFixture {
 		ctx:              context.Background(),
 		wgInterface:      &mocWGIface{},
 		statusRecorder:   recorder,
+		dnsMuxMap:        make(registeredHandlerMap),
 		selectedRoutes:   func() route.HAMap { return fx.selected },
 		activeRoutes:     func() route.HAMap { return fx.active },
 		warningDelayBase: defaultWarningDelayBase,
 	}
-	fx.server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: fx.stub, priority: PriorityUpstream}}
+	fx.server.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: fx.stub, priority: PriorityUpstream}
 
 	fx.server.mux.Lock()
 	fx.server.updateNSGroupStates([]*nbdns.NameServerGroup{fx.group})
@@ -2018,6 +2395,7 @@ func TestProjection_OverlayAddrNoRouteDelaysWarning(t *testing.T) {
 		ctx:              context.Background(),
 		wgInterface:      &mocWGIface{},
 		statusRecorder:   recorder,
+		dnsMuxMap:        make(registeredHandlerMap),
 		selectedRoutes:   func() route.HAMap { return nil },
 		activeRoutes:     func() route.HAMap { return nil },
 		warningDelayBase: 50 * time.Millisecond,
@@ -2029,7 +2407,7 @@ func TestProjection_OverlayAddrNoRouteDelaysWarning(t *testing.T) {
 	stub := &healthStubHandler{health: map[netip.AddrPort]UpstreamHealth{
 		overlayPeer: {LastFail: time.Now(), LastErr: "timeout"},
 	}}
-	server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: stub, priority: PriorityUpstream}}
+	server.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: stub, priority: PriorityUpstream}
 
 	server.mux.Lock()
 	server.updateNSGroupStates([]*nbdns.NameServerGroup{group})
@@ -2066,6 +2444,7 @@ func TestProjection_StopClearsHealthState(t *testing.T) {
 		service:           NewServiceViaMemory(wgIface),
 		hostManager:       &noopHostConfigurator{},
 		extraDomains:      map[domain.Domain]int{},
+		dnsMuxMap:         make(registeredHandlerMap),
 		statusRecorder:    peer.NewRecorder("mgm"),
 		selectedRoutes:    func() route.HAMap { return nil },
 		activeRoutes:      func() route.HAMap { return nil },
@@ -2080,7 +2459,7 @@ func TestProjection_StopClearsHealthState(t *testing.T) {
 		NameServers: []nbdns.NameServer{{IP: srv.Addr(), NSType: nbdns.UDPNameServerType, Port: int(srv.Port())}},
 	}
 	stub := &healthStubHandler{health: map[netip.AddrPort]UpstreamHealth{srv: {LastOk: time.Now()}}}
-	server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: stub, priority: PriorityUpstream}}
+	server.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: stub, priority: PriorityUpstream}
 
 	server.mux.Lock()
 	server.updateNSGroupStates([]*nbdns.NameServerGroup{group})
@@ -2105,32 +2484,6 @@ func TestProjection_StopClearsHealthState(t *testing.T) {
 // rule 3: startup failures while the peer is handshaking, then the peer
 // comes up and a query succeeds before the grace window elapses. No
 // warning should ever have fired, and no recovery either.
-func TestWarningDelayBaseFromEnv(t *testing.T) {
-	tests := []struct {
-		name string
-		set  bool
-		val  string
-		want time.Duration
-	}{
-		{name: "unset uses default", set: false, want: defaultWarningDelayBase},
-		{name: "valid override", set: true, val: "90s", want: 90 * time.Second},
-		{name: "valid minutes", set: true, val: "2m", want: 2 * time.Minute},
-		{name: "invalid falls back", set: true, val: "notaduration", want: defaultWarningDelayBase},
-		{name: "zero falls back", set: true, val: "0s", want: defaultWarningDelayBase},
-		{name: "negative falls back", set: true, val: "-30s", want: defaultWarningDelayBase},
-	}
-
-	for _, tc := range tests {
-		t.Run(tc.name, func(t *testing.T) {
-			t.Setenv(envWarningDelay, tc.val)
-			if !tc.set {
-				os.Unsetenv(envWarningDelay)
-			}
-			assert.Equal(t, tc.want, warningDelayBaseFromEnv(), "grace window base")
-		})
-	}
-}
-
 func TestProjection_OverlayRecoversDuringGrace(t *testing.T) {
 	fx := newProjTestFixture(t)
 	fx.server.warningDelayBase = 200 * time.Millisecond
@@ -2242,6 +2595,7 @@ func TestProjection_MixedGroupEmitsImmediately(t *testing.T) {
 	server := &DefaultServer{
 		ctx:              context.Background(),
 		statusRecorder:   recorder,
+		dnsMuxMap:        make(registeredHandlerMap),
 		selectedRoutes:   func() route.HAMap { return overlayMap },
 		activeRoutes:     func() route.HAMap { return nil },
 		warningDelayBase: time.Hour,
@@ -2259,7 +2613,7 @@ func TestProjection_MixedGroupEmitsImmediately(t *testing.T) {
 			overlay: {LastFail: time.Now(), LastErr: "timeout"},
 		},
 	}
-	server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: stub, priority: PriorityUpstream}}
+	server.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: stub, priority: PriorityUpstream}
 
 	server.mux.Lock()
 	server.updateNSGroupStates([]*nbdns.NameServerGroup{group})
@@ -2286,6 +2640,7 @@ func TestDNSLoopPrevention(t *testing.T) {
 		localResolver: local.NewResolver(),
 		handlerChain:  NewHandlerChain(),
 		hostManager:   &noopHostConfigurator{},
+		dnsMuxMap:     make(registeredHandlerMap),
 	}
 
 	tests := []struct {

client/internal/dns/upstream.go

@@ -443,32 +443,29 @@ func (u *upstreamResolverBase) queryUpstream(parentCtx context.Context, r *dns.M
 		return raceResult{}, &upstreamFailure{upstream: upstream, reason: "no response"}
 	}
 
-	// A valid response means the upstream is reachable, whatever the Rcode.
-	u.markUpstreamOk(upstream)
-
 	proto := ""
 	if upstreamProto != nil {
 		proto = upstreamProto.protocol
 	}
 
 	if rm.Rcode == dns.RcodeServerFailure || rm.Rcode == dns.RcodeRefused {
-		// SERVFAIL and REFUSED are per-question outcomes (DNSSEC-bogus names,
-		// refused zones, transient recursion errors), not reachability
-		// problems: fail over for a better answer but keep the upstream healthy.
 		if code, ok := nonRetryableEDE(rm); ok {
 			if !hadEdns {
-				resutil.StripOPT(rm)
+				stripOPT(rm)
 			}
+			u.markUpstreamOk(upstream)
 			return raceResult{msg: rm, upstream: upstream, protocol: proto, ede: edeName(code)}, nil
 		}
 		reason := dns.RcodeToString[rm.Rcode]
+		u.markUpstreamFail(upstream, reason)
 		return raceResult{}, &upstreamFailure{upstream: upstream, reason: reason}
 	}
 
 	if !hadEdns {
-		resutil.StripOPT(rm)
+		stripOPT(rm)
 	}
 
+	u.markUpstreamOk(upstream)
 	return raceResult{msg: rm, upstream: upstream, protocol: proto}, nil
 }
 
@@ -523,6 +520,22 @@ func upstreamUDPSize() uint16 {
 	return dns.MinMsgSize
 }
 
+// stripOPT removes any OPT pseudo-RRs from the response's Extra section so
+// the response complies with RFC 6891 when the client did not advertise EDNS0.
+func stripOPT(rm *dns.Msg) {
+	if len(rm.Extra) == 0 {
+		return
+	}
+	out := rm.Extra[:0]
+	for _, rr := range rm.Extra {
+		if _, ok := rr.(*dns.OPT); ok {
+			continue
+		}
+		out = append(out, rr)
+	}
+	rm.Extra = out
+}
+
 func (u *upstreamResolverBase) handleUpstreamError(err error, upstream netip.AddrPort, startTime time.Time) *upstreamFailure {
 	if !errors.Is(err, context.DeadlineExceeded) && !isTimeout(err) {
 		return &upstreamFailure{upstream: upstream, reason: err.Error()}

client/internal/dns/upstream_test.go

@@ -517,78 +517,6 @@ func TestUpstreamResolver_HealthTracking(t *testing.T) {
 	assert.NotContains(t, health, bad, "sibling upstream should not be queried when primary answers")
 }
 
-// TestUpstreamResolver_HealthTracking_ResponseMeansReachable verifies that an
-// upstream which answers with SERVFAIL or REFUSED is recorded as healthy:
-// those are per-question outcomes from a reachable server and must not mark
-// the upstream unhealthy. Only transport failures (timeouts) do.
-func TestUpstreamResolver_HealthTracking_ResponseMeansReachable(t *testing.T) {
-	a := netip.MustParseAddrPort("192.0.2.10:53")
-	b := netip.MustParseAddrPort("192.0.2.11:53")
-	timeoutErr := &net.OpError{Op: "read", Err: fmt.Errorf("i/o timeout")}
-
-	tests := []struct {
-		name        string
-		respA       mockUpstreamResponse
-		respB       mockUpstreamResponse
-		wantHealthy bool
-	}{
-		{
-			name:        "both SERVFAIL are reachable",
-			respA:       mockUpstreamResponse{msg: buildMockResponse(dns.RcodeServerFailure, "")},
-			respB:       mockUpstreamResponse{msg: buildMockResponse(dns.RcodeServerFailure, "")},
-			wantHealthy: true,
-		},
-		{
-			name:        "both REFUSED are reachable",
-			respA:       mockUpstreamResponse{msg: buildMockResponse(dns.RcodeRefused, "")},
-			respB:       mockUpstreamResponse{msg: buildMockResponse(dns.RcodeRefused, "")},
-			wantHealthy: true,
-		},
-		{
-			name:        "timeout marks unhealthy",
-			respA:       mockUpstreamResponse{err: timeoutErr},
-			respB:       mockUpstreamResponse{err: timeoutErr},
-			wantHealthy: false,
-		},
-	}
-
-	for _, tc := range tests {
-		t.Run(tc.name, func(t *testing.T) {
-			mockClient := &mockUpstreamResolverPerServer{
-				responses: map[string]mockUpstreamResponse{
-					a.String(): tc.respA,
-					b.String(): tc.respB,
-				},
-				rtt: time.Millisecond,
-			}
-
-			ctx, cancel := context.WithCancel(context.Background())
-			defer cancel()
-
-			resolver := &upstreamResolverBase{
-				ctx:             ctx,
-				upstreamClient:  mockClient,
-				upstreamTimeout: UpstreamTimeout,
-			}
-			resolver.addRace([]netip.AddrPort{a, b})
-
-			responseWriter := &test.MockResponseWriter{WriteMsgFunc: func(m *dns.Msg) error { return nil }}
-			resolver.ServeDNS(responseWriter, new(dns.Msg).SetQuestion("example.com.", dns.TypeA))
-
-			health := resolver.UpstreamHealth()
-			require.Contains(t, health, a, "primary upstream should have a health record")
-			if tc.wantHealthy {
-				assert.False(t, health[a].LastOk.IsZero(), "responding upstream should have LastOk set")
-				assert.True(t, health[a].LastFail.IsZero(), "responding upstream should not be marked failed")
-				assert.Empty(t, health[a].LastErr, "responding upstream should have no error")
-			} else {
-				assert.False(t, health[a].LastFail.IsZero(), "timed-out upstream should be marked failed")
-				assert.NotEmpty(t, health[a].LastErr, "timed-out upstream should record an error")
-			}
-		})
-	}
-}
-
 func TestFormatFailures(t *testing.T) {
 	testCases := []struct {
 		name     string
@@ -985,6 +913,19 @@ func TestEDEName(t *testing.T) {
 	assert.Equal(t, "EDE 9999", edeName(9999), "unknown code falls back to numeric")
 }
 
+func TestStripOPT(t *testing.T) {
+	rm := &dns.Msg{
+		Extra: []dns.RR{
+			&dns.OPT{Hdr: dns.RR_Header{Name: ".", Rrtype: dns.TypeOPT}},
+			&dns.A{Hdr: dns.RR_Header{Name: "x.", Rrtype: dns.TypeA}, A: net.IPv4(1, 2, 3, 4)},
+		},
+	}
+	stripOPT(rm)
+	assert.Len(t, rm.Extra, 1, "OPT should be removed, A kept")
+	_, isOPT := rm.Extra[0].(*dns.OPT)
+	assert.False(t, isOPT, "remaining record must not be OPT")
+}
+
 func TestUpstreamResolver_NonRetryableEDEShortCircuits(t *testing.T) {
 	upstream1 := netip.MustParseAddrPort("192.0.2.1:53")
 	upstream2 := netip.MustParseAddrPort("192.0.2.2:53")

client/internal/dnsfwd/forwarder.go

@@ -26,23 +26,8 @@ import (
 const errResolveFailed = "failed to resolve query for domain=%s: %v"
 const upstreamTimeout = 15 * time.Second
 
-// EDE info codes the forwarder emits on upstream failures so the querying
-// client can see the reason without inspecting this peer's logs. They live in
-// the RFC 8914 Private Use range (49152-65535); the Go resolver never exposes a
-// real upstream EDE here, so these cannot collide with a genuine code.
-const (
-	edeNetbirdUpstreamTimeout uint16 = 49152
-	edeNetbirdUpstreamFailure uint16 = 49153
-)
-
 type resolver interface {
 	LookupNetIP(ctx context.Context, network, host string) ([]netip.Addr, error)
-	LookupMX(ctx context.Context, name string) ([]*net.MX, error)
-	LookupTXT(ctx context.Context, name string) ([]string, error)
-	LookupNS(ctx context.Context, name string) ([]*net.NS, error)
-	LookupSRV(ctx context.Context, service, proto, name string) (string, []*net.SRV, error)
-	LookupCNAME(ctx context.Context, host string) (string, error)
-	LookupAddr(ctx context.Context, addr string) ([]string, error)
 }
 
 type firewaller interface {
@@ -216,6 +201,12 @@ func (f *DNSForwarder) handleDNSQuery(logger *log.Entry, w dns.ResponseWriter, q
 		qname, dns.TypeToString[question.Qtype], dns.ClassToString[question.Qclass])
 
 	resp := query.SetReply(query)
+	network := resutil.NetworkForQtype(question.Qtype)
+	if network == "" {
+		resp.Rcode = dns.RcodeNotImplemented
+		f.writeResponse(logger, w, resp, qname, startTime)
+		return
+	}
 
 	mostSpecificResId, matchingEntries := f.getMatchingEntries(strings.TrimSuffix(qname, "."))
 	if mostSpecificResId == "" {
@@ -227,46 +218,9 @@ func (f *DNSForwarder) handleDNSQuery(logger *log.Entry, w dns.ResponseWriter, q
 	ctx, cancel := context.WithTimeout(context.Background(), upstreamTimeout)
 	defer cancel()
 
-	reqHasEdns := query.IsEdns0() != nil
-
-	switch question.Qtype {
-	case dns.TypeA, dns.TypeAAAA:
-		f.handleAddressQuery(ctx, logger, w, resp, mostSpecificResId, matchingEntries, reqHasEdns, startTime)
-	case dns.TypeMX, dns.TypeTXT, dns.TypeNS, dns.TypeSRV, dns.TypeCNAME, dns.TypePTR:
-		f.handleRecordQuery(ctx, logger, w, resp, startTime)
-	default:
-		// The domain is routed here, so any other type is answered NODATA
-		// (NOERROR, empty answer) rather than falling back to a resolver that
-		// would poison the name with NXDOMAIN. The Extended DNS Error lets a
-		// client tell this capability-driven NODATA apart from an
-		// authoritative one. The OPT pseudo-record must not appear unless the
-		// query advertised EDNS0.
-		if reqHasEdns {
-			attachEDE(resp, dns.ExtendedErrorCodeNotSupported, "netbird forwarder: unsupported query type")
-		}
-		f.writeResponse(logger, w, resp, qname, startTime)
-	}
-}
-
-// handleAddressQuery resolves A/AAAA queries, programs the firewall sets and
-// resolved-IP state, and caches the answer for resilience on upstream failure.
-func (f *DNSForwarder) handleAddressQuery(
-	ctx context.Context,
-	logger *log.Entry,
-	w dns.ResponseWriter,
-	resp *dns.Msg,
-	mostSpecificResId route.ResID,
-	matchingEntries []*ForwarderEntry,
-	reqHasEdns bool,
-	startTime time.Time,
-) {
-	question := resp.Question[0]
-	qname := strings.ToLower(question.Name)
-
-	network := resutil.NetworkForQtype(question.Qtype)
 	result := resutil.LookupIP(ctx, f.resolver, network, qname, question.Qtype)
 	if result.Err != nil {
-		f.handleDNSError(ctx, logger, w, question, resp, qname, result, reqHasEdns, startTime)
+		f.handleDNSError(ctx, logger, w, question, resp, qname, result, startTime)
 		return
 	}
 
@@ -277,25 +231,6 @@ func (f *DNSForwarder) handleAddressQuery(
 	f.writeResponse(logger, w, resp, qname, startTime)
 }
 
-// handleRecordQuery resolves non-address record types (MX, TXT, NS, SRV,
-// CNAME, PTR) through the host resolver. Missing records are answered NODATA so
-// the routed name is never poisoned with NXDOMAIN.
-func (f *DNSForwarder) handleRecordQuery(
-	ctx context.Context,
-	logger *log.Entry,
-	w dns.ResponseWriter,
-	resp *dns.Msg,
-	startTime time.Time,
-) {
-	question := resp.Question[0]
-	qname := strings.ToLower(question.Name)
-
-	records, rcode := resutil.LookupRecords(ctx, f.resolver, qname, question.Qtype, f.ttl)
-	resp.Rcode = rcode
-	resp.Answer = append(resp.Answer, records...)
-	f.writeResponse(logger, w, resp, qname, startTime)
-}
-
 func (f *DNSForwarder) writeResponse(logger *log.Entry, w dns.ResponseWriter, resp *dns.Msg, qname string, startTime time.Time) {
 	if err := w.WriteMsg(resp); err != nil {
 		logger.Errorf("failed to write DNS response: %v", err)
@@ -398,7 +333,6 @@ func (f *DNSForwarder) handleDNSError(
 	resp *dns.Msg,
 	domain string,
 	result resutil.LookupResult,
-	reqHasEdns bool,
 	startTime time.Time,
 ) {
 	qType := question.Qtype
@@ -440,10 +374,6 @@ func (f *DNSForwarder) handleDNSError(
 		logger.Warnf(errResolveFailed, domain, result.Err)
 	}
 
-	if reqHasEdns {
-		attachEDE(resp, edeCodeFor(dnsErr), edeText(dnsErr))
-	}
-
 	f.writeResponse(logger, w, resp, domain, startTime)
 }
 
@@ -484,33 +414,3 @@ func (f *DNSForwarder) getMatchingEntries(domain string) (route.ResID, []*Forwar
 
 	return selectedResId, matches
 }
-
-// edeCodeFor maps an upstream lookup error to the NetBird EDE info code.
-func edeCodeFor(dnsErr *net.DNSError) uint16 {
-	if dnsErr != nil && dnsErr.IsTimeout {
-		return edeNetbirdUpstreamTimeout
-	}
-	return edeNetbirdUpstreamFailure
-}
-
-// edeText builds the EDE extra-text describing the class of upstream failure.
-// It deliberately omits the upstream server address, which may be an internal
-// resolver and is exposed to any client permitted to use the route; the full
-// detail stays in the forwarder's local log.
-func edeText(dnsErr *net.DNSError) string {
-	if dnsErr != nil && dnsErr.IsTimeout {
-		return "netbird forwarder: upstream timeout"
-	}
-	return "netbird forwarder: upstream failure"
-}
-
-// attachEDE adds an Extended DNS Error (RFC 8914) option to the response,
-// creating the OPT pseudo-record if the response does not already carry one.
-func attachEDE(resp *dns.Msg, code uint16, text string) {
-	opt := resp.IsEdns0()
-	if opt == nil {
-		resp.SetEdns0(dns.DefaultMsgSize, false)
-		opt = resp.IsEdns0()
-	}
-	opt.Option = append(opt.Option, &dns.EDNS0_EDE{InfoCode: code, ExtraText: text})
-}

client/internal/dnsfwd/forwarder_test.go

@@ -16,7 +16,6 @@ import (
 	"github.com/stretchr/testify/require"
 
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
-	"github.com/netbirdio/netbird/client/internal/dns/resutil"
 	"github.com/netbirdio/netbird/client/internal/dns/test"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/route"
@@ -133,41 +132,6 @@ func (m *MockResolver) LookupNetIP(ctx context.Context, network, host string) ([
 	return args.Get(0).([]netip.Addr), args.Error(1)
 }
 
-func (m *MockResolver) LookupMX(ctx context.Context, name string) ([]*net.MX, error) {
-	args := m.Called(ctx, name)
-	recs, _ := args.Get(0).([]*net.MX)
-	return recs, args.Error(1)
-}
-
-func (m *MockResolver) LookupTXT(ctx context.Context, name string) ([]string, error) {
-	args := m.Called(ctx, name)
-	recs, _ := args.Get(0).([]string)
-	return recs, args.Error(1)
-}
-
-func (m *MockResolver) LookupNS(ctx context.Context, name string) ([]*net.NS, error) {
-	args := m.Called(ctx, name)
-	recs, _ := args.Get(0).([]*net.NS)
-	return recs, args.Error(1)
-}
-
-func (m *MockResolver) LookupSRV(ctx context.Context, service, proto, name string) (string, []*net.SRV, error) {
-	args := m.Called(ctx, service, proto, name)
-	recs, _ := args.Get(1).([]*net.SRV)
-	return args.String(0), recs, args.Error(2)
-}
-
-func (m *MockResolver) LookupCNAME(ctx context.Context, host string) (string, error) {
-	args := m.Called(ctx, host)
-	return args.String(0), args.Error(1)
-}
-
-func (m *MockResolver) LookupAddr(ctx context.Context, addr string) ([]string, error) {
-	args := m.Called(ctx, addr)
-	recs, _ := args.Get(0).([]string)
-	return recs, args.Error(1)
-}
-
 func TestDNSForwarder_SubdomainAccessLogic(t *testing.T) {
 	tests := []struct {
 		name             string
@@ -580,15 +544,12 @@ func TestDNSForwarder_MultipleIPsInSingleUpdate(t *testing.T) {
 }
 
 func TestDNSForwarder_ResponseCodes(t *testing.T) {
-	// A type with no net.Resolver Lookup method (CAA) must answer NODATA
-	// (NOERROR, empty) rather than NXDOMAIN/NOTIMP to avoid poisoning the name.
 	tests := []struct {
 		name         string
 		queryType    uint16
 		queryDomain  string
 		configured   string
 		expectedCode int
-		expectEDE    bool
 		description  string
 	}{
 		{
@@ -600,13 +561,28 @@ func TestDNSForwarder_ResponseCodes(t *testing.T) {
 			description:  "RFC compliant REFUSED for unauthorized queries",
 		},
 		{
-			name:         "unsupported query type returns NODATA",
-			queryType:    dns.TypeCAA,
+			name:         "unsupported query type returns NOTIMP",
+			queryType:    dns.TypeMX,
 			queryDomain:  "example.com",
 			configured:   "example.com",
-			expectedCode: dns.RcodeSuccess,
-			expectEDE:    true,
-			description:  "Unsupported types answer NODATA, not NXDOMAIN/NOTIMP",
+			expectedCode: dns.RcodeNotImplemented,
+			description:  "RFC compliant NOTIMP for unsupported types",
+		},
+		{
+			name:         "CNAME query returns NOTIMP",
+			queryType:    dns.TypeCNAME,
+			queryDomain:  "example.com",
+			configured:   "example.com",
+			expectedCode: dns.RcodeNotImplemented,
+			description:  "CNAME queries not supported",
+		},
+		{
+			name:         "TXT query returns NOTIMP",
+			queryType:    dns.TypeTXT,
+			queryDomain:  "example.com",
+			configured:   "example.com",
+			expectedCode: dns.RcodeNotImplemented,
+			description:  "TXT queries not supported",
 		},
 	}
 
@@ -622,7 +598,6 @@ func TestDNSForwarder_ResponseCodes(t *testing.T) {
 
 			query := &dns.Msg{}
 			query.SetQuestion(dns.Fqdn(tt.queryDomain), tt.queryType)
-			query.SetEdns0(dns.DefaultMsgSize, false)
 
 			// Capture the written response
 			var writtenResp *dns.Msg
@@ -638,288 +613,6 @@ func TestDNSForwarder_ResponseCodes(t *testing.T) {
 			// Check the response written to the writer
 			require.NotNil(t, writtenResp, "Expected response to be written")
 			assert.Equal(t, tt.expectedCode, writtenResp.Rcode, tt.description)
-			assert.Empty(t, writtenResp.Answer, "Non-address response should carry no answers")
-
-			if tt.expectEDE {
-				require.NotNil(t, writtenResp.IsEdns0(), "EDNS0 client should get an OPT in the reply")
-				assert.True(t, hasEDE(writtenResp, dns.ExtendedErrorCodeNotSupported),
-					"unsupported type NODATA should carry EDE Not Supported")
-			}
-		})
-	}
-}
-
-func hasEDE(m *dns.Msg, code uint16) bool {
-	opt := m.IsEdns0()
-	if opt == nil {
-		return false
-	}
-	for _, o := range opt.Option {
-		if ede, ok := o.(*dns.EDNS0_EDE); ok && ede.InfoCode == code {
-			return true
-		}
-	}
-	return false
-}
-
-func TestDNSForwarder_RecordQueries(t *testing.T) {
-	notFound := &net.DNSError{IsNotFound: true, Name: "example.com"}
-
-	t.Run("MX records are forwarded", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
-
-		mockResolver.On("LookupMX", mock.Anything, "example.com.").
-			Return([]*net.MX{{Host: "mail.example.com.", Pref: 10}}, nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeMX)
-		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		require.Len(t, resp.Answer, 1)
-		mx, ok := resp.Answer[0].(*dns.MX)
-		require.True(t, ok, "answer should be an MX record")
-		assert.Equal(t, uint16(10), mx.Preference)
-		assert.Equal(t, "mail.example.com.", mx.Mx)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("missing MX is NODATA not NXDOMAIN", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
-
-		// A not-found cannot prove the name is absent (it may exist with only
-		// other record types), so it must answer NODATA, never NXDOMAIN.
-		mockResolver.On("LookupMX", mock.Anything, "example.com.").
-			Return(nil, notFound).Once()
-
-		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeMX)
-		assert.Equal(t, dns.RcodeSuccess, resp.Rcode, "missing record must be NODATA")
-		assert.Empty(t, resp.Answer)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("NS records are forwarded", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
-
-		mockResolver.On("LookupNS", mock.Anything, "example.com.").
-			Return([]*net.NS{{Host: "ns1.example.com."}}, nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeNS)
-		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		require.Len(t, resp.Answer, 1)
-		ns, ok := resp.Answer[0].(*dns.NS)
-		require.True(t, ok, "answer should be an NS record")
-		assert.Equal(t, "ns1.example.com.", ns.Ns)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("missing NS is NODATA", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
-
-		mockResolver.On("LookupNS", mock.Anything, "example.com.").
-			Return(nil, notFound).Once()
-
-		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeNS)
-		assert.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		assert.Empty(t, resp.Answer)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("SRV records are forwarded", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "_sip._tcp.example.com")
-
-		mockResolver.On("LookupSRV", mock.Anything, "", "", "_sip._tcp.example.com.").
-			Return("", []*net.SRV{{Target: "sip.example.com.", Port: 5060, Priority: 10, Weight: 5}}, nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "_sip._tcp.example.com", dns.TypeSRV)
-		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		require.Len(t, resp.Answer, 1)
-		srv, ok := resp.Answer[0].(*dns.SRV)
-		require.True(t, ok, "answer should be an SRV record")
-		assert.Equal(t, "sip.example.com.", srv.Target)
-		assert.Equal(t, uint16(5060), srv.Port)
-		assert.Equal(t, uint16(10), srv.Priority)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("missing SRV is NODATA", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "_sip._tcp.example.com")
-
-		mockResolver.On("LookupSRV", mock.Anything, "", "", "_sip._tcp.example.com.").
-			Return("", nil, notFound).Once()
-
-		resp := runRecordQuery(t, forwarder, "_sip._tcp.example.com", dns.TypeSRV)
-		assert.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		assert.Empty(t, resp.Answer)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("TXT records are forwarded", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
-
-		mockResolver.On("LookupTXT", mock.Anything, "example.com.").
-			Return([]string{"v=spf1 -all"}, nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeTXT)
-		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		require.Len(t, resp.Answer, 1)
-		txt, ok := resp.Answer[0].(*dns.TXT)
-		require.True(t, ok, "answer should be a TXT record")
-		assert.Equal(t, []string{"v=spf1 -all"}, txt.Txt)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("CNAME record is forwarded", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "www.example.com")
-
-		mockResolver.On("LookupCNAME", mock.Anything, "www.example.com.").
-			Return("target.example.com.", nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "www.example.com", dns.TypeCNAME)
-		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		require.Len(t, resp.Answer, 1)
-		cname, ok := resp.Answer[0].(*dns.CNAME)
-		require.True(t, ok, "answer should be a CNAME record")
-		assert.Equal(t, "target.example.com.", cname.Target)
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("CNAME equal to the name is NODATA", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
-
-		// No CNAME exists: LookupCNAME echoes the queried name back.
-		mockResolver.On("LookupCNAME", mock.Anything, "example.com.").
-			Return("example.com.", nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeCNAME)
-		assert.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		assert.Empty(t, resp.Answer, "self-referential CNAME means no CNAME record")
-		mockResolver.AssertExpectations(t)
-	})
-
-	t.Run("PTR record is forwarded", func(t *testing.T) {
-		mockResolver := &MockResolver{}
-		forwarder := newRecordTestForwarder(t, mockResolver, "*.in-addr.arpa")
-
-		// The reverse name is parsed back to the address LookupAddr expects.
-		mockResolver.On("LookupAddr", mock.Anything, "1.2.3.4").
-			Return([]string{"host.example.com."}, nil).Once()
-
-		resp := runRecordQuery(t, forwarder, "4.3.2.1.in-addr.arpa", dns.TypePTR)
-		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
-		require.Len(t, resp.Answer, 1)
-		ptr, ok := resp.Answer[0].(*dns.PTR)
-		require.True(t, ok, "answer should be a PTR record")
-		assert.Equal(t, "host.example.com.", ptr.Ptr)
-		mockResolver.AssertExpectations(t)
-	})
-}
-
-func newRecordTestForwarder(t *testing.T, r resolver, configured string) *DNSForwarder {
-	t.Helper()
-	forwarder := NewDNSForwarder(netip.MustParseAddrPort("127.0.0.1:0"), 300, nil, &peer.Status{}, nil)
-	forwarder.resolver = r
-
-	d, err := domain.FromString(configured)
-	require.NoError(t, err)
-	forwarder.UpdateDomains([]*ForwarderEntry{{Domain: d, ResID: "test-res"}})
-	return forwarder
-}
-
-func runRecordQuery(t *testing.T, forwarder *DNSForwarder, qname string, qtype uint16) *dns.Msg {
-	t.Helper()
-	query := &dns.Msg{}
-	query.SetQuestion(dns.Fqdn(qname), qtype)
-
-	mockWriter := &test.MockResponseWriter{}
-	forwarder.handleDNSQuery(log.NewEntry(log.StandardLogger()), mockWriter, query, time.Now())
-
-	resp := mockWriter.GetLastResponse()
-	require.NotNil(t, resp, "expected response to be written")
-	return resp
-}
-
-func TestDNSForwarder_UpstreamFailureEDE(t *testing.T) {
-	tests := []struct {
-		name        string
-		lookupErr   error
-		reqEdns     bool
-		wantEDE     bool
-		wantCode    uint16
-		wantTextHas string
-	}{
-		{
-			name:        "timeout with edns0",
-			lookupErr:   &net.DNSError{Err: "i/o timeout", Server: "10.0.0.53:53", IsTimeout: true},
-			reqEdns:     true,
-			wantEDE:     true,
-			wantCode:    edeNetbirdUpstreamTimeout,
-			wantTextHas: "netbird forwarder: upstream timeout",
-		},
-		{
-			name:        "server failure with edns0",
-			lookupErr:   &net.DNSError{Err: "server misbehaving", Server: "10.0.0.53:53"},
-			reqEdns:     true,
-			wantEDE:     true,
-			wantCode:    edeNetbirdUpstreamFailure,
-			wantTextHas: "netbird forwarder: upstream failure",
-		},
-		{
-			name:      "no edns0 in request omits ede",
-			lookupErr: &net.DNSError{Err: "server misbehaving", Server: "10.0.0.53:53"},
-			reqEdns:   false,
-			wantEDE:   false,
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			mockResolver := &MockResolver{}
-			forwarder := NewDNSForwarder(netip.MustParseAddrPort("127.0.0.1:0"), 300, nil, &peer.Status{}, nil)
-			forwarder.resolver = mockResolver
-
-			d, err := domain.FromString("example.com")
-			require.NoError(t, err)
-			forwarder.UpdateDomains([]*ForwarderEntry{{Domain: d, ResID: "test-res"}})
-
-			mockResolver.On("LookupNetIP", mock.Anything, "ip4", "example.com.").
-				Return([]netip.Addr(nil), tt.lookupErr).Once()
-
-			query := &dns.Msg{}
-			query.SetQuestion("example.com.", dns.TypeA)
-			if tt.reqEdns {
-				query.SetEdns0(dns.DefaultMsgSize, false)
-			}
-
-			var writtenResp *dns.Msg
-			mockWriter := &test.MockResponseWriter{
-				WriteMsgFunc: func(m *dns.Msg) error {
-					writtenResp = m
-					return nil
-				},
-			}
-
-			forwarder.handleDNSQuery(log.NewEntry(log.StandardLogger()), mockWriter, query, time.Now())
-			mockResolver.AssertExpectations(t)
-
-			require.NotNil(t, writtenResp, "expected a response")
-			assert.Equal(t, dns.RcodeServerFailure, writtenResp.Rcode, "upstream failure must be SERVFAIL")
-
-			ede, ok := resutil.ExtractEDE(writtenResp)
-			if !tt.wantEDE {
-				assert.False(t, ok, "response must not carry EDE")
-				return
-			}
-			require.True(t, ok, "response must carry EDE")
-			assert.Equal(t, tt.wantCode, ede.InfoCode, "EDE info code")
-			assert.Contains(t, ede.ExtraText, tt.wantTextHas, "EDE extra-text")
-			assert.NotContains(t, ede.ExtraText, "10.0.0.53", "must not leak upstream server address")
 		})
 	}
 }

client/internal/engine.go

@@ -82,18 +82,10 @@ const (
 	PeerConnectionTimeoutMax = 45000 // ms
 	PeerConnectionTimeoutMin = 30000 // ms
 	disableAutoUpdate        = "disabled"
-
-	// systemInfoTimeout bounds how long the sync loop waits for system info / posture
-	// check gathering. The gathering runs uncancellable system calls (process scan,
-	// exec, os.Stat); without this bound a single stuck call freezes handleSync, and
-	// thus syncMsgMux, for as long as the call hangs (observed multi-minute freezes).
-	systemInfoTimeout = 15 * time.Second
 )
 
 var ErrResetConnection = fmt.Errorf("reset connection")
 
-var ErrEngineAlreadyStarted = errors.New("engine already started")
-
 type EngineConfig struct {
 	WgPort      int
 	WgIfaceName string
@@ -207,8 +199,6 @@ type Engine struct {
 	ctx    context.Context
 	cancel context.CancelFunc
 
-	started bool
-
 	wgInterface WGIface
 
 	udpMux *udpmux.UniversalUDPMuxDefault
@@ -216,12 +206,6 @@ type Engine struct {
 	// networkSerial is the latest CurrentSerial (state ID) of the network sent by the Management service
 	networkSerial uint64
 
-	// forwardingRules holds the ingress forward rules applied for the current target.
-	// Wholesale sections (incl. forward rules) run only on the first pass of a target;
-	// it is stashed here so the final, peer-converged pass can build the lazy-connection
-	// exclude list without recomputing them on every bounded peer pass.
-	forwardingRules []firewallManager.ForwardRule
-
 	networkMonitor *networkmonitor.NetworkMonitor
 
 	sshServer sshServer
@@ -275,6 +259,20 @@ type Engine struct {
 	jobExecutorWG sync.WaitGroup
 
 	exposeManager *expose.Manager
+
+	sessionWatcher sessionDeadlineWatcher
+}
+
+// sessionDeadlineWatcher is the engine-facing surface of the SSO session
+// expiry watcher. The concrete implementation (sessionwatch.Watcher) is wired
+// in via newSessionWatcher, which is build-tagged so the js/wasm build links a
+// no-op stub instead of pulling the full sessionwatch package (and its timer
+// machinery) into the binary — the wasm client never runs the engine's
+// session-warning flow.
+type sessionDeadlineWatcher interface {
+	Update(deadline time.Time) error
+	Dismiss()
+	Close()
 }
 
 // Peer is an instance of the Connection Peer
@@ -295,15 +293,9 @@ func NewEngine(
 	services EngineServices,
 	mobileDep MobileDependency,
 ) *Engine {
-	// The engine is single-use: a fresh instance is built per connection
-	// cycle (see Client.run), so the run context is created once here rather
-	// than in Start.
-	ctx, cancel := context.WithCancel(clientCtx)
 	engine := &Engine{
 		clientCtx:          clientCtx,
 		clientCancel:       clientCancel,
-		ctx:                ctx,
-		cancel:             cancel,
 		signal:             services.SignalClient,
 		signaler:           peer.NewSignaler(services.SignalClient, config.WgPrivateKey),
 		mgmClient:          services.MgmClient,
@@ -325,6 +317,17 @@ func NewEngine(
 		updateManager:      services.UpdateManager,
 		syncStoreDir:       config.StateDir,
 	}
+	// sessionWatcher keeps the SubscribeStatus consumers in sync with the
+	// session expiry deadline. Deadline-change ticks come for free via
+	// Status.SetSessionExpiresAt; the watcher exists to push a wake-up at
+	// T-WarningLead and T-FinalWarningLead so the UI repaints the remaining
+	// time / warning state even when nothing else changed, and to publish
+	// two SystemEvents (the warning composition lives in sessionwatch so
+	// the wire format stays owned by one package):
+	//   - T-WarningLead   → interactive "Extend now / Dismiss" notification
+	//   - T-FinalWarningLead → auto-opened SessionAboutToExpire dialog,
+	//     suppressed when the user dismissed the earlier warning
+	engine.sessionWatcher = newSessionWatcher(engine.statusRecorder)
 
 	log.Infof("I am: %s", config.WgPrivateKey.PublicKey().String())
 	return engine
@@ -336,34 +339,8 @@ func (e *Engine) Stop() error {
 		log.Debugf("tried stopping engine that is nil")
 		return nil
 	}
-	e.cancel()
 	e.syncMsgMux.Lock()
 
-	e.stopLocked()
-
-	e.syncMsgMux.Unlock()
-
-	timeout := e.calculateShutdownTimeout()
-	log.Debugf("waiting for goroutines to finish with timeout: %v", timeout)
-	shutdownCtx, cancel := context.WithTimeout(context.Background(), timeout)
-	defer cancel()
-
-	if err := waitWithContext(shutdownCtx, &e.shutdownWg); err != nil {
-		log.Warnf("shutdown timeout exceeded after %v, some goroutines may still be running", timeout)
-	}
-
-	log.Infof("stopped Netbird Engine")
-
-	return nil
-}
-
-// stopLocked tears down everything Start may have brought up, in the order
-// teardown requires (DNS before the interface goes down, flow manager after).
-// The caller must hold syncMsgMux. It is shared by Stop and by Start's failure
-// path, so a partially-initialized engine is cleaned up the same way; every
-// step is nil-guarded. It does not wait on shutdownWg — the caller does that
-// after releasing the lock, since the goroutines also take syncMsgMux.
-func (e *Engine) stopLocked() {
 	if e.connMgr != nil {
 		e.connMgr.Close()
 	}
@@ -391,6 +368,10 @@ func (e *Engine) stopLocked() {
 		e.srWatcher.Close()
 	}
 
+	if e.sessionWatcher != nil {
+		e.sessionWatcher.Close()
+	}
+
 	if e.updateManager != nil {
 		e.updateManager.SetDownloadOnly()
 	}
@@ -414,6 +395,10 @@ func (e *Engine) stopLocked() {
 	// so dbus and friends don't complain because of a missing interface
 	e.stopDNSServer()
 
+	if e.cancel != nil {
+		e.cancel()
+	}
+
 	e.jobExecutorWG.Wait() // block until job goroutines finish
 
 	e.close()
@@ -432,6 +417,21 @@ func (e *Engine) stopLocked() {
 	if err := e.stateManager.PersistState(context.Background()); err != nil {
 		log.Errorf("failed to persist state: %v", err)
 	}
+
+	e.syncMsgMux.Unlock()
+
+	timeout := e.calculateShutdownTimeout()
+	log.Debugf("waiting for goroutines to finish with timeout: %v", timeout)
+	shutdownCtx, cancel := context.WithTimeout(context.Background(), timeout)
+	defer cancel()
+
+	if err := waitWithContext(shutdownCtx, &e.shutdownWg); err != nil {
+		log.Warnf("shutdown timeout exceeded after %v, some goroutines may still be running", timeout)
+	}
+
+	log.Infof("stopped Netbird Engine")
+
+	return nil
 }
 
 // calculateShutdownTimeout returns shutdown timeout: 10s base + 100ms per peer, capped at 30s.
@@ -469,38 +469,18 @@ func waitWithContext(ctx context.Context, wg *sync.WaitGroup) error {
 // Start creates a new WireGuard tunnel interface and listens to events from Signal and Management services
 // Connections to remote peers are not established here.
 // However, they will be established once an event with a list of peers to connect to will be received from Management Service
-func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL) (err error) {
+func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL) error {
 	e.syncMsgMux.Lock()
 	defer e.syncMsgMux.Unlock()
 
-	// The engine is single-use. Reject a duplicate start and a start on an
-	// already-stopped engine (run context cancelled).
-	if e.started {
-		return ErrEngineAlreadyStarted
-	}
-
-	if ctxErr := e.ctx.Err(); ctxErr != nil {
-		return fmt.Errorf("engine already stopped: %w", ctxErr)
-	}
-
-	e.started = true
-
-	// Tear down any partially-initialized state on a failed start. Cancel the
-	// run context first so goroutines started before the failure (connMgr,
-	// srWatcher, monitors) unwind, then stopLocked mirrors Stop's teardown (we
-	// already hold syncMsgMux), cleaning up route/DNS/flow/state managers too,
-	// not just what close() covers.
-	defer func() {
-		if err != nil {
-			e.cancel()
-			e.stopLocked()
-		}
-	}()
-
-	if err = iface.ValidateMTU(e.config.MTU); err != nil {
+	if err := iface.ValidateMTU(e.config.MTU); err != nil {
 		return fmt.Errorf("invalid MTU configuration: %w", err)
 	}
 
+	if e.cancel != nil {
+		e.cancel()
+	}
+	e.ctx, e.cancel = context.WithCancel(e.clientCtx)
 	e.exposeManager = expose.NewManager(e.ctx, e.mgmClient)
 
 	wgIface, err := e.newWgIface()
@@ -534,11 +514,13 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 
 	initialRoutes, dnsConfig, dnsFeatureFlag, err := e.readInitialSettings()
 	if err != nil {
+		e.close()
 		return fmt.Errorf("read initial settings: %w", err)
 	}
 
 	dnsServer, err := e.newDnsServer(dnsConfig)
 	if err != nil {
+		e.close()
 		return fmt.Errorf("create dns server: %w", err)
 	}
 	e.dnsServer = dnsServer
@@ -573,6 +555,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 
 	if err = e.wgInterfaceCreate(); err != nil {
 		log.Errorf("failed creating tunnel interface %s: [%s]", e.config.WgIfaceName, err.Error())
+		e.close()
 		return fmt.Errorf("create wg interface: %w", err)
 	}
 
@@ -581,6 +564,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 	}
 
 	if err := e.createFirewall(); err != nil {
+		e.close()
 		return err
 	}
 
@@ -592,6 +576,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 	e.udpMux, err = e.wgInterface.Up()
 	if err != nil {
 		log.Errorf("failed to pull up wgInterface [%s]: %s", e.wgInterface.Name(), err.Error())
+		e.close()
 		return fmt.Errorf("up wg interface: %w", err)
 	}
 
@@ -616,7 +601,9 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 		e.acl = acl.NewDefaultManager(e.firewall)
 	}
 
-	if err := e.dnsServer.Initialize(); err != nil {
+	err = e.dnsServer.Initialize()
+	if err != nil {
+		e.close()
 		return fmt.Errorf("initialize dns server: %w", err)
 	}
 
@@ -628,9 +615,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 	e.srWatcher = guard.NewSRWatcher(e.signal, e.relayManager, e.mobileDep.IFaceDiscover, iceCfg)
 	e.srWatcher.Start(peer.IsForceRelayed())
 
-	if err = e.receiveSignalEvents(); err != nil {
-		return err
-	}
+	e.receiveSignalEvents()
 	e.receiveManagementEvents()
 	e.receiveJobEvents()
 
@@ -682,6 +667,7 @@ func (e *Engine) createFirewall() error {
 
 func (e *Engine) initFirewall() error {
 	if err := e.routeManager.SetFirewall(e.firewall); err != nil {
+		e.close()
 		return fmt.Errorf("set firewall: %w", err)
 	}
 
@@ -774,15 +760,7 @@ func (e *Engine) blockLanAccess() {
 
 // modifyPeers updates peers that have been modified (e.g. IP address has been changed).
 // It closes the existing connection, removes it from the peerConns map, and creates a new one.
-// maxPeersPerSyncPass is the default per-pass cap on how many peers each of
-// removePeers/modifyPeers/addNewPeers applies, so syncMsgMux is held only for a
-// batch at a time and other subsystems can interleave between passes. It is
-// passed in (not read globally) so tests can exercise the multi-pass path.
-const maxPeersPerSyncPass = 300
-
-// modifyPeers re-applies up to maxBatch changed peers per call. It returns true
-// when more changed peers remained than the cap, so the caller re-runs.
-func (e *Engine) modifyPeers(peersUpdate []*mgmProto.RemotePeerConfig, maxBatch int) (bool, error) {
+func (e *Engine) modifyPeers(peersUpdate []*mgmProto.RemotePeerConfig) error {
 
 	// first, check if peers have been modified
 	var modified []*mgmProto.RemotePeerConfig
@@ -812,32 +790,26 @@ func (e *Engine) modifyPeers(peersUpdate []*mgmProto.RemotePeerConfig, maxBatch
 		}
 	}
 
-	more := false
-	if len(modified) > maxBatch {
-		modified = modified[:maxBatch]
-		more = true
-	}
-
 	// second, close all modified connections and remove them from the state map
 	for _, p := range modified {
-		if err := e.removePeer(p.GetWgPubKey()); err != nil {
-			return false, err
+		err := e.removePeer(p.GetWgPubKey())
+		if err != nil {
+			return err
 		}
 	}
 	// third, add the peer connections again
 	for _, p := range modified {
-		if err := e.addNewPeer(p); err != nil {
-			return false, err
+		err := e.addNewPeer(p)
+		if err != nil {
+			return err
 		}
 	}
-	return more, nil
+	return nil
 }
 
 // removePeers finds and removes peers that do not exist anymore in the network map received from the Management Service.
 // It also removes peers that have been modified (e.g. change of IP address). They will be added again in addPeers method.
-// removePeers removes up to maxBatch peers per call. It returns true when more
-// peers remained to remove than the cap, so the caller re-runs.
-func (e *Engine) removePeers(peersUpdate []*mgmProto.RemotePeerConfig, maxBatch int) (bool, error) {
+func (e *Engine) removePeers(peersUpdate []*mgmProto.RemotePeerConfig) error {
 	newPeers := make([]string, 0, len(peersUpdate))
 	for _, p := range peersUpdate {
 		newPeers = append(newPeers, p.GetWgPubKey())
@@ -845,19 +817,14 @@ func (e *Engine) removePeers(peersUpdate []*mgmProto.RemotePeerConfig, maxBatch
 
 	toRemove := util.SliceDiff(e.peerStore.PeersPubKey(), newPeers)
 
-	more := false
-	if len(toRemove) > maxBatch {
-		toRemove = toRemove[:maxBatch]
-		more = true
-	}
-
 	for _, p := range toRemove {
-		if err := e.removePeer(p); err != nil {
-			return false, err
+		err := e.removePeer(p)
+		if err != nil {
+			return err
 		}
 		log.Infof("removed peer %s", p)
 	}
-	return more, nil
+	return nil
 }
 
 func (e *Engine) removeAllPeers() error {
@@ -926,38 +893,29 @@ func (e *Engine) handleAutoUpdateVersion(autoUpdateSettings *mgmProto.AutoUpdate
 	e.updateManager.SetVersion(autoUpdateSettings.Version, autoUpdateSettings.AlwaysUpdate)
 }
 
-// phase times a sync sub-phase: it returns a function that records the elapsed
-// duration when called. Starting the timer at the call site keeps inter-phase
-// glue code out of the measurement.
-func (e *Engine) phase(name string) func() {
-	start := time.Now()
-	return func() {
-		e.clientMetrics.RecordSyncPhase(e.ctx, name, time.Since(start))
-	}
-}
-
-// applySyncPass applies one bounded pass of the sync update under syncMsgMux and
-// returns true if more peers remained than the per-pass cap. It is driven by the
-// mapStateManager, which re-invokes it (releasing the lock between passes) until
-// the update is fully applied.
-func (e *Engine) applySyncPass(update *mgmProto.SyncResponse, firstPass bool) (bool, error) {
+func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
+	started := time.Now()
+	defer func() {
+		duration := time.Since(started)
+		log.Infof("sync finished in %s", duration)
+		e.clientMetrics.RecordSyncDuration(e.ctx, duration)
+	}()
 	e.syncMsgMux.Lock()
 	defer e.syncMsgMux.Unlock()
 
 	// Check context INSIDE lock to ensure atomicity with shutdown
 	if e.ctx.Err() != nil {
-		return false, e.ctx.Err()
+		return e.ctx.Err()
 	}
 
+	e.ApplySessionDeadline(update.GetSessionExpiresAt())
+
 	if update.NetworkMap != nil && update.NetworkMap.PeerConfig != nil {
 		e.handleAutoUpdateVersion(update.NetworkMap.PeerConfig.AutoUpdate)
 	}
 
-	done := e.phase("netbird_config")
-	err := e.updateNetbirdConfig(update.GetNetbirdConfig())
-	done()
-	if err != nil {
-		return false, err
+	if err := e.updateNetbirdConfig(update.GetNetbirdConfig()); err != nil {
+		return err
 	}
 
 	// Posture checks are bound to the network map presence:
@@ -967,25 +925,23 @@ func (e *Engine) applySyncPass(update *mgmProto.SyncResponse, firstPass bool) (b
 	//                                        leave the previously applied checks untouched
 	nm := update.GetNetworkMap()
 	if nm == nil {
-		return false, nil
+		return nil
 	}
 
-	done = e.phase("checks")
-	err = e.updateChecksIfNew(update.Checks)
-	done()
-	if err != nil {
-		return false, err
+	if err := e.updateChecksIfNew(update.Checks); err != nil {
+		return err
 	}
 
+	e.persistSyncResponse(update)
+
 	// only apply new changes and ignore old ones
-	more, err := e.updateNetworkMap(nm, maxPeersPerSyncPass, firstPass)
-	if err != nil {
-		return false, err
+	if err := e.updateNetworkMap(nm); err != nil {
+		return err
 	}
 
 	e.statusRecorder.PublishEvent(cProto.SystemEvent_INFO, cProto.SystemEvent_SYSTEM, "Network map updated", "", nil)
 
-	return more, nil
+	return nil
 }
 
 // updateNetbirdConfig applies the management-provided NetBird configuration:
@@ -1031,13 +987,6 @@ func (e *Engine) updateNetbirdConfig(wCfg *mgmProto.NetbirdConfig) error {
 // (not syncMsgMux) is held for the whole Set so the store cannot be cleared (disabled /
 // engine close) mid-call and have this write resurrect a file that was just removed.
 func (e *Engine) persistSyncResponse(update *mgmProto.SyncResponse) {
-	// Only persist updates that carry a network map. Config-only updates (e.g. relay
-	// token rotation, STUN/TURN) have a nil NetworkMap; persisting them would overwrite
-	// the last full map on disk and break restore-on-restart.
-	if update.GetNetworkMap() == nil {
-		return
-	}
-
 	e.syncRespMux.RLock()
 	defer e.syncRespMux.RUnlock()
 
@@ -1117,22 +1066,11 @@ func (e *Engine) updateChecksIfNew(checks []*mgmProto.Checks) error {
 	}
 	e.checks = checks
 
-	info, ok := system.GetInfoWithChecksTimeout(e.ctx, systemInfoTimeout, checks, e.overlayAddresses()...)
-	if !ok {
-		// Gathering timed out; skip the meta sync this cycle rather than blocking the
-		// sync loop (and syncMsgMux) on a stuck system call. A later sync will retry.
-		return nil
+	info, err := system.GetInfoWithChecks(e.ctx, checks)
+	if err != nil {
+		log.Warnf("failed to get system info with checks: %v", err)
+		info = system.GetInfo(e.ctx)
 	}
-	e.applyInfoFlags(info)
-
-	if err := e.mgmClient.SyncMeta(info); err != nil {
-		return fmt.Errorf("could not sync meta: error %s", err)
-	}
-	return nil
-}
-
-// applyInfoFlags sets the engine's config-derived feature flags on the gathered system info.
-func (e *Engine) applyInfoFlags(info *system.Info) {
 	info.SetFlags(
 		e.config.RosenpassEnabled,
 		e.config.RosenpassPermissive,
@@ -1151,20 +1089,12 @@ func (e *Engine) applyInfoFlags(info *system.Info) {
 		e.config.EnableSSHRemotePortForwarding,
 		e.config.DisableSSHAuth,
 	)
-}
 
-// overlayAddresses returns our own WireGuard overlay address (v4 and v6) so it
-// can be excluded from the reported network addresses; the interface coming and
-// going otherwise churns the peer meta on the management server.
-func (e *Engine) overlayAddresses() []netip.Addr {
-	var ips []netip.Addr
-	if e.config.WgAddr.IP.IsValid() {
-		ips = append(ips, e.config.WgAddr.IP)
+	if err := e.mgmClient.SyncMeta(info); err != nil {
+		log.Errorf("could not sync meta: error %s", err)
+		return err
 	}
-	if e.config.WgAddr.HasIPv6() {
-		ips = append(ips, e.config.WgAddr.IPv6)
-	}
-	return ips
+	return nil
 }
 
 func (e *Engine) updateConfig(conf *mgmProto.PeerConfig) error {
@@ -1279,7 +1209,7 @@ func (e *Engine) handleBundle(params *mgmProto.BundleParameters) (*mgmProto.JobR
 		ClientMetrics:  e.clientMetrics,
 		DaemonVersion:  version.NetbirdVersion(),
 		RefreshStatus: func() {
-			e.RunHealthProbes(true)
+			e.RunHealthProbes(e.ctx, true)
 		},
 	}
 
@@ -1310,32 +1240,31 @@ func (e *Engine) receiveManagementEvents() {
 	e.shutdownWg.Add(1)
 	go func() {
 		defer e.shutdownWg.Done()
-		info, ok := system.GetInfoWithChecksTimeout(e.ctx, systemInfoTimeout, e.checks, e.overlayAddresses()...)
-		if !ok {
-			// Gathering timed out; connect the stream with base info so management
-			// connectivity still comes up rather than blocking here.
+		info, err := system.GetInfoWithChecks(e.ctx, e.checks)
+		if err != nil {
+			log.Warnf("failed to get system info with checks: %v", err)
 			info = system.GetInfo(e.ctx)
 		}
-		e.applyInfoFlags(info)
+		info.SetFlags(
+			e.config.RosenpassEnabled,
+			e.config.RosenpassPermissive,
+			&e.config.ServerSSHAllowed,
+			e.config.DisableClientRoutes,
+			e.config.DisableServerRoutes,
+			e.config.DisableDNS,
+			e.config.DisableFirewall,
+			e.config.BlockLANAccess,
+			e.config.BlockInbound,
+			e.config.DisableIPv6,
+			e.config.LazyConnectionEnabled,
+			e.config.EnableSSHRoot,
+			e.config.EnableSSHSFTP,
+			e.config.EnableSSHLocalPortForwarding,
+			e.config.EnableSSHRemotePortForwarding,
+			e.config.DisableSSHAuth,
+		)
 
-		// The map-state manager converges the latest update in the background in
-		// bounded passes; the stream callback only hands it the newest target.
-		persist := func(u *mgmProto.SyncResponse) {
-			done := e.phase("persist")
-			e.persistSyncResponse(u)
-			done()
-		}
-		manager := newMapStateManager(e.applySyncPass, persist, func(d time.Duration) {
-			log.Infof("sync finished in %s", d)
-			e.clientMetrics.RecordSyncDuration(e.ctx, d)
-		})
-		e.shutdownWg.Add(1)
-		go func() {
-			defer e.shutdownWg.Done()
-			manager.run(e.ctx)
-		}()
-
-		err := e.mgmClient.Sync(e.ctx, info, manager.SetTarget)
+		err = e.mgmClient.Sync(e.ctx, info, e.handleSync)
 		if err != nil {
 			// happens if management is unavailable for a long time.
 			// We want to cancel the operation of the whole client
@@ -1386,107 +1315,21 @@ func (e *Engine) updateTURNs(turns []*mgmProto.ProtectedHostConfig) error {
 	return nil
 }
 
-// updateNetworkMap applies the wholesale parts (config, routes, ACL, DNS) in full
-// and up to maxBatch peers per phase. It returns true when more peers remained
-// than the cap, so the caller re-runs until convergence.
-func (e *Engine) updateNetworkMap(networkMap *mgmProto.NetworkMap, maxBatch int, firstPass bool) (bool, error) {
+func (e *Engine) updateNetworkMap(networkMap *mgmProto.NetworkMap) error {
 	// intentionally leave it before checking serial because for now it can happen that peer IP changed but serial didn't
 	if networkMap.GetPeerConfig() != nil {
 		err := e.updateConfig(networkMap.GetPeerConfig())
 		if err != nil {
-			return false, err
+			return err
 		}
 	}
 
 	serial := networkMap.GetSerial()
 	if e.networkSerial > serial {
 		log.Debugf("received outdated NetworkMap with serial %d, ignoring", serial)
-		return false, nil
+		return nil
 	}
 
-	// Wholesale sections (firewall/ACL, DNS, routes, forward rules) are applied
-	// up-front and only once per target: they are cheap, local, idempotent and must
-	// be in place before peers come up (fail-closed). On the bounded re-runs that only
-	// drain the remaining peer batches they are skipped — the applied forward rules are
-	// reused from e.forwardingRules for the lazy-exclude finalize.
-	if firstPass {
-		e.applyWholesale(networkMap, serial)
-	}
-
-	log.Debugf("got peers update from Management Service, total peers to connect to = %d", len(networkMap.GetRemotePeers()))
-
-	doneOffline := e.phase("offline_peers")
-	e.updateOfflinePeers(networkMap.GetOfflinePeers())
-	doneOffline()
-
-	// Filter out own peer from the remote peers list
-	localPubKey := e.config.WgPrivateKey.PublicKey().String()
-	remotePeers := make([]*mgmProto.RemotePeerConfig, 0, len(networkMap.GetRemotePeers()))
-	for _, p := range networkMap.GetRemotePeers() {
-		if p.GetWgPubKey() != localPubKey {
-			remotePeers = append(remotePeers, p)
-		}
-	}
-
-	// No special case for cleanup: when management signals RemotePeersIsEmpty (e.g. our
-	// peer was deleted), remotePeers is already empty, so the bounded diff below removes
-	// every peer in batches — same path as a normal update, no unbounded removeAllPeers
-	// held under syncMsgMux in one shot.
-	doneRemoved := e.phase("removed_peers")
-	removeMore, err := e.removePeers(remotePeers, maxBatch)
-	doneRemoved()
-	if err != nil {
-		return false, err
-	}
-
-	doneModified := e.phase("modified_peers")
-	modifyMore, err := e.modifyPeers(remotePeers, maxBatch)
-	doneModified()
-	if err != nil {
-		return false, err
-	}
-
-	doneAdded := e.phase("added_peers")
-	addMore, err := e.addNewPeers(remotePeers, maxBatch)
-	doneAdded()
-	if err != nil {
-		return false, err
-	}
-
-	// needMore signals the caller to re-run when a peer phase hit its per-pass cap.
-	needMore := removeMore || modifyMore || addMore
-
-	e.statusRecorder.FinishPeerListModifications()
-
-	e.updatePeerSSHHostKeys(remotePeers)
-
-	if err := e.updateSSHClientConfig(remotePeers); err != nil {
-		log.Warnf("failed to update SSH client config: %v", err)
-	}
-
-	e.updateSSHServerAuth(networkMap.GetSshAuth())
-
-	// Set the exclude list only once peers have fully converged (this pass added
-	// the last batch). It needs all target peers present in the store, and
-	// ExcludePeer has replace-semantics — a partial set mid-convergence would be wrong.
-	if !needMore {
-		doneLazy := e.phase("lazy_exclude")
-		excludedLazyPeers := e.toExcludedLazyPeers(e.forwardingRules, remotePeers)
-		e.connMgr.SetExcludeList(e.ctx, excludedLazyPeers)
-		doneLazy()
-	}
-
-	e.networkSerial = serial
-
-	return needMore, nil
-}
-
-// applyWholesale applies the cheap, local, idempotent map sections — lazy feature
-// flag, firewall/legacy management, DNS, routes, ACL filtering, DNS forwarder and
-// ingress forward rules — that must be in place before peers come up. It runs once
-// per target (first pass only); the resulting forward rules are stashed in
-// e.forwardingRules for the lazy-exclude finalize on the peer-converged pass.
-func (e *Engine) applyWholesale(networkMap *mgmProto.NetworkMap, serial uint64) {
 	if err := e.connMgr.UpdatedRemoteFeatureFlag(e.ctx, networkMap.GetPeerConfig().GetLazyConnectionEnabled()); err != nil {
 		log.Errorf("failed to update lazy connection feature flag: %v", err)
 	}
@@ -1514,16 +1357,13 @@ func (e *Engine) applyWholesale(networkMap *mgmProto.NetworkMap, serial uint64)
 
 	dnsConfig := toDNSConfig(protoDNSConfig, e.wgInterface.Address())
 
-	done := e.phase("dns_server")
 	if err := e.dnsServer.UpdateDNSServer(serial, dnsConfig); err != nil {
 		log.Errorf("failed to update dns server, err: %v", err)
 	}
-	done()
 
 	e.routeManager.SetDNSForwarderPort(dnsConfig.ForwarderPort)
 
 	// apply routes first, route related actions might depend on routing being enabled
-	done = e.phase("routes_classify")
 	routes := toRoutes(networkMap.GetRoutes())
 	serverRoutes, clientRoutes := e.routeManager.ClassifyRoutes(routes)
 
@@ -1532,34 +1372,79 @@ func (e *Engine) applyWholesale(networkMap *mgmProto.NetworkMap, serial uint64)
 		e.connMgr.UpdateRouteHAMap(clientRoutes)
 		log.Debugf("updated lazy connection manager with %d HA groups", len(clientRoutes))
 	}
-	done()
 
-	done = e.phase("routes_apply")
 	dnsRouteFeatureFlag := toDNSFeatureFlag(networkMap)
 	if err := e.routeManager.UpdateRoutes(serial, serverRoutes, clientRoutes, dnsRouteFeatureFlag); err != nil {
 		log.Errorf("failed to update routes: %v", err)
 	}
-	done()
 
-	done = e.phase("filtering")
 	if e.acl != nil {
 		e.acl.ApplyFiltering(networkMap, dnsRouteFeatureFlag)
 	}
-	done()
 
-	done = e.phase("dns_forwarder")
 	fwdEntries := toRouteDomains(e.config.WgPrivateKey.PublicKey().String(), routes)
 	e.updateDNSForwarder(dnsRouteFeatureFlag, fwdEntries)
-	done()
 
 	// Ingress forward rules
-	done = e.phase("forward_rules")
 	forwardingRules, err := e.updateForwardRules(networkMap.GetForwardingRules())
 	if err != nil {
 		log.Errorf("failed to update forward rules, err: %v", err)
 	}
-	done()
-	e.forwardingRules = forwardingRules
+
+	log.Debugf("got peers update from Management Service, total peers to connect to = %d", len(networkMap.GetRemotePeers()))
+
+	e.updateOfflinePeers(networkMap.GetOfflinePeers())
+
+	// Filter out own peer from the remote peers list
+	localPubKey := e.config.WgPrivateKey.PublicKey().String()
+	remotePeers := make([]*mgmProto.RemotePeerConfig, 0, len(networkMap.GetRemotePeers()))
+	for _, p := range networkMap.GetRemotePeers() {
+		if p.GetWgPubKey() != localPubKey {
+			remotePeers = append(remotePeers, p)
+		}
+	}
+
+	// cleanup request, most likely our peer has been deleted
+	if networkMap.GetRemotePeersIsEmpty() {
+		err := e.removeAllPeers()
+		e.statusRecorder.FinishPeerListModifications()
+		if err != nil {
+			return err
+		}
+	} else {
+		err := e.removePeers(remotePeers)
+		if err != nil {
+			return err
+		}
+
+		err = e.modifyPeers(remotePeers)
+		if err != nil {
+			return err
+		}
+
+		err = e.addNewPeers(remotePeers)
+		if err != nil {
+			return err
+		}
+
+		e.statusRecorder.FinishPeerListModifications()
+
+		e.updatePeerSSHHostKeys(remotePeers)
+
+		if err := e.updateSSHClientConfig(remotePeers); err != nil {
+			log.Warnf("failed to update SSH client config: %v", err)
+		}
+
+		e.updateSSHServerAuth(networkMap.GetSshAuth())
+	}
+
+	// must set the exclude list after the peers are added. Without it the manager can not figure out the peers parameters from the store
+	excludedLazyPeers := e.toExcludedLazyPeers(forwardingRules, remotePeers)
+	e.connMgr.SetExcludeList(e.ctx, excludedLazyPeers)
+
+	e.networkSerial = serial
+
+	return nil
 }
 
 func toDNSFeatureFlag(networkMap *mgmProto.NetworkMap) bool {
@@ -1739,23 +1624,14 @@ func addrToString(addr netip.Addr) string {
 }
 
 // addNewPeers adds peers that were not know before but arrived from the Management service with the update
-// addNewPeers adds up to maxBatch not-yet-present peers per call. It returns true
-// when more new peers remained than the cap, so the caller re-runs.
-func (e *Engine) addNewPeers(peersUpdate []*mgmProto.RemotePeerConfig, maxBatch int) (bool, error) {
-	added := 0
+func (e *Engine) addNewPeers(peersUpdate []*mgmProto.RemotePeerConfig) error {
 	for _, p := range peersUpdate {
-		if _, ok := e.peerStore.PeerConn(p.GetWgPubKey()); ok {
-			continue // already present (cheap skip), does not count toward the cap
+		err := e.addNewPeer(p)
+		if err != nil {
+			return err
 		}
-		if added >= maxBatch {
-			return true, nil // at least one more new peer remains
-		}
-		if err := e.addNewPeer(p); err != nil {
-			return false, err
-		}
-		added++
 	}
-	return false, nil
+	return nil
 }
 
 // addNewPeer add peer if connection doesn't exist
@@ -1853,7 +1729,7 @@ func (e *Engine) createPeerConn(pubKey string, allowedIPs []netip.Prefix, agentV
 }
 
 // receiveSignalEvents connects to the Signal Service event stream to negotiate connection with remote peers
-func (e *Engine) receiveSignalEvents() error {
+func (e *Engine) receiveSignalEvents() {
 	e.shutdownWg.Add(1)
 	go func() {
 		defer e.shutdownWg.Done()
@@ -1924,12 +1800,7 @@ func (e *Engine) receiveSignalEvents() error {
 		}
 	}()
 
-	// todo: consider to remove this blocker. I do not see benefit to block the Start operations
-	e.signal.WaitStreamConnected(e.ctx)
-	if err := e.ctx.Err(); err != nil {
-		return fmt.Errorf("wait for signal stream: %w", err)
-	}
-	return nil
+	e.signal.WaitStreamConnected()
 }
 
 func (e *Engine) parseNATExternalIPMappings() []string {
@@ -2241,7 +2112,20 @@ func (e *Engine) getRosenpassAddr() string {
 
 // RunHealthProbes executes health checks for Signal, Management, Relay, and WireGuard services
 // and updates the status recorder with the latest states.
-func (e *Engine) RunHealthProbes(waitForResult bool) bool {
+//
+// ctx scopes the (potentially slow) STUN/TURN probing: a caller that gives up —
+// e.g. a Status RPC whose client disconnected — cancels its ctx and the probe
+// returns instead of running to its per-component timeout. The engine's own
+// lifetime ctx still applies independently, so an engine shutdown aborts the
+// probe even if the caller's ctx is context.Background().
+func (e *Engine) RunHealthProbes(ctx context.Context, waitForResult bool) bool {
+	// Tie the caller's ctx to the engine lifetime: either cancelling aborts
+	// the probe below.
+	ctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+	stop := context.AfterFunc(e.ctx, cancel)
+	defer stop()
+
 	e.syncMsgMux.Lock()
 
 	signalHealthy := e.signal.IsHealthy()
@@ -2264,9 +2148,9 @@ func (e *Engine) RunHealthProbes(waitForResult bool) bool {
 	if runtime.GOOS != "js" {
 		var results []relay.ProbeResult
 		if waitForResult {
-			results = e.probeStunTurn.ProbeAllWaitResult(e.ctx, stuns, turns)
+			results = e.probeStunTurn.ProbeAllWaitResult(ctx, stuns, turns)
 		} else {
-			results = e.probeStunTurn.ProbeAll(e.ctx, stuns, turns)
+			results = e.probeStunTurn.ProbeAll(ctx, stuns, turns)
 		}
 		e.statusRecorder.UpdateRelayStates(results)
 

client/internal/engine_authsession.go

@@ -0,0 +1,108 @@
+package internal
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"time"
+
+	log "github.com/sirupsen/logrus"
+	"google.golang.org/protobuf/types/known/timestamppb"
+
+	"github.com/netbirdio/netbird/client/internal/auth/sessionwatch"
+	cProto "github.com/netbirdio/netbird/client/proto"
+	"github.com/netbirdio/netbird/client/system"
+)
+
+// ApplySessionDeadline propagates the absolute SSO session deadline carried on
+// LoginResponse / SyncResponse to both the watcher (for the edge-triggered
+// warning) and the status recorder (for the SubscribeStatus / Status RPC
+// snapshot the UI consumes).
+//
+// The wire field is 3-state:
+//   - nil                        → snapshot carries no info; keep the
+//     previously-anchored deadline (no-op)
+//   - explicit zero (s=0, n=0)   → peer is not SSO-registered or expiry is
+//     disabled; clear both sinks
+//   - valid timestamp            → new deadline; arm watcher, expose on
+//     status recorder
+//
+// Deadline sanity-checks live in sessionwatch.Watcher.Update. Any rejected
+// value is treated as a clear on both sinks: the alternative — leaving the
+// previously-known deadline in place — risks the UI confidently displaying
+// a stale "expires in X" while the server has actually invalidated it.
+func (e *Engine) ApplySessionDeadline(ts *timestamppb.Timestamp) {
+	if ts == nil {
+		return
+	}
+	var deadline time.Time
+	// Explicit zero (seconds=0 AND nanos=0) is the sentinel for "disabled".
+	// Everything else flows through Watcher.Update, whose sanity-checks
+	// reject out-of-range / pre-epoch / far-future / too-stale values and
+	// clear on rejection.
+	if ts.GetSeconds() != 0 || ts.GetNanos() != 0 {
+		deadline = ts.AsTime().UTC()
+	}
+	if e.sessionWatcher == nil {
+		return
+	}
+	// Watcher.Update owns the propagation to the status recorder (the
+	// SubscribeStatus / Status snapshot the UI reads): a set writes the
+	// deadline, a clear or a sanity-check rejection writes the zero value.
+	// Keeping a single writer is what stops the recorder from drifting out
+	// of sync with the warning timers.
+	if err := e.sessionWatcher.Update(deadline); err != nil {
+		log.Errorf("auth session deadline rejected: %v, clearing", err)
+		e.statusRecorder.PublishEvent(
+			cProto.SystemEvent_ERROR,
+			cProto.SystemEvent_AUTHENTICATION,
+			"session deadline rejected",
+			"",
+			map[string]string{sessionwatch.MetaSessionDeadlineRejected: err.Error()},
+		)
+	}
+}
+
+// DismissSessionWarning records the user's "Dismiss" click on the
+// T-WarningLead interactive notification and suppresses the upcoming
+// T-FinalWarningLead fallback for the current deadline. No-op when the
+// watcher is not running or holds no deadline.
+func (e *Engine) DismissSessionWarning() {
+	if e.sessionWatcher == nil {
+		return
+	}
+	e.sessionWatcher.Dismiss()
+}
+
+// ExtendAuthSession asks the management server to refresh the SSO session
+// expiry deadline using the supplied JWT, then mirrors the new deadline into
+// the daemon's state. The tunnel is untouched; no resync, no reconnect.
+//
+// Returns the new absolute UTC deadline (or zero time when the server
+// reports the peer is not eligible for extension).
+func (e *Engine) ExtendAuthSession(ctx context.Context, jwtToken string) (time.Time, error) {
+	if jwtToken == "" {
+		return time.Time{}, errors.New("jwt token is required")
+	}
+	if e.mgmClient == nil {
+		return time.Time{}, errors.New("management client is not initialised")
+	}
+
+	info, err := system.GetInfoWithChecks(ctx, e.checks)
+	if err != nil {
+		log.Warnf("failed to collect system info for session extend: %v", err)
+		info = system.GetInfo(ctx)
+	}
+
+	resp, err := e.mgmClient.ExtendAuthSession(info, jwtToken)
+	if err != nil {
+		return time.Time{}, fmt.Errorf("extend auth session on management: %w", err)
+	}
+
+	e.ApplySessionDeadline(resp.GetSessionExpiresAt())
+
+	if resp.GetSessionExpiresAt().IsValid() {
+		return resp.GetSessionExpiresAt().AsTime().UTC(), nil
+	}
+	return time.Time{}, nil
+}

client/internal/engine_lazyconn.go

@@ -0,0 +1,19 @@
+package internal
+
+import (
+	"errors"
+)
+
+// SetLazyConnEnabled applies a local lazy connection override to the running
+// engine. It pins the setting like an env/CLI flag, so a later management sync
+// cannot override it. syncMsgMux guards ConnMgr, which is not thread-safe.
+func (e *Engine) SetLazyConnEnabled(enabled bool) error {
+	e.syncMsgMux.Lock()
+	defer e.syncMsgMux.Unlock()
+
+	if e.connMgr == nil {
+		return errors.New("connection manager is not initialised")
+	}
+
+	return e.connMgr.SetLocalLazyConn(e.ctx, enabled)
+}

client/internal/engine_privileged_test.go

@@ -1,565 +0,0 @@
-//go:build privileged
-
-package internal
-
-import (
-	"context"
-	"fmt"
-	"net"
-	"runtime"
-	"strings"
-	"sync"
-	"testing"
-	"time"
-
-	"github.com/golang/mock/gomock"
-	"github.com/google/uuid"
-	log "github.com/sirupsen/logrus"
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-	"go.opentelemetry.io/otel"
-	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
-	"google.golang.org/grpc"
-	"google.golang.org/grpc/keepalive"
-
-	"github.com/netbirdio/netbird/client/iface"
-	"github.com/netbirdio/netbird/client/iface/device"
-	"github.com/netbirdio/netbird/client/iface/wgaddr"
-	"github.com/netbirdio/netbird/client/internal/dns"
-	"github.com/netbirdio/netbird/client/internal/peer"
-	nbssh "github.com/netbirdio/netbird/client/ssh"
-	"github.com/netbirdio/netbird/client/system"
-	nbdns "github.com/netbirdio/netbird/dns"
-	"github.com/netbirdio/netbird/management/internals/controllers/network_map/controller"
-	"github.com/netbirdio/netbird/management/internals/controllers/network_map/update_channel"
-	"github.com/netbirdio/netbird/management/internals/modules/peers"
-	"github.com/netbirdio/netbird/management/internals/modules/peers/ephemeral/manager"
-	"github.com/netbirdio/netbird/management/internals/server/config"
-	nbgrpc "github.com/netbirdio/netbird/management/internals/shared/grpc"
-	"github.com/netbirdio/netbird/management/server"
-	"github.com/netbirdio/netbird/management/server/activity"
-	nbcache "github.com/netbirdio/netbird/management/server/cache"
-	"github.com/netbirdio/netbird/management/server/groups"
-	"github.com/netbirdio/netbird/management/server/integrations/integrated_validator/validator"
-	"github.com/netbirdio/netbird/management/server/integrations/port_forwarding"
-	"github.com/netbirdio/netbird/management/server/job"
-	"github.com/netbirdio/netbird/management/server/permissions"
-	"github.com/netbirdio/netbird/management/server/settings"
-	"github.com/netbirdio/netbird/management/server/store"
-	"github.com/netbirdio/netbird/management/server/telemetry"
-	"github.com/netbirdio/netbird/management/server/types"
-	mgmt "github.com/netbirdio/netbird/shared/management/client"
-	mgmtProto "github.com/netbirdio/netbird/shared/management/proto"
-	relayClient "github.com/netbirdio/netbird/shared/relay/client"
-	signal "github.com/netbirdio/netbird/shared/signal/client"
-	"github.com/netbirdio/netbird/shared/signal/proto"
-	signalServer "github.com/netbirdio/netbird/signal/server"
-	"github.com/netbirdio/netbird/util"
-)
-
-func TestEngine_SSH(t *testing.T) {
-	key, err := wgtypes.GeneratePrivateKey()
-	if err != nil {
-		t.Fatal(err)
-		return
-	}
-
-	sshKey, err := nbssh.GeneratePrivateKey(nbssh.ED25519)
-	if err != nil {
-		t.Fatal(err)
-		return
-	}
-
-	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
-	defer cancel()
-
-	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
-	engine := NewEngine(
-		ctx, cancel,
-		&EngineConfig{
-			WgIfaceName:      "utun101",
-			WgAddr:           wgaddr.MustParseWGAddress("100.64.0.1/24"),
-			WgPrivateKey:     key,
-			WgPort:           33100,
-			ServerSSHAllowed: true,
-			MTU:              iface.DefaultMTU,
-			SSHKey:           sshKey,
-		},
-		EngineServices{
-			SignalClient:   &signal.MockClient{},
-			MgmClient:      &mgmt.MockClient{},
-			RelayManager:   relayMgr,
-			StatusRecorder: peer.NewRecorder("https://mgm"),
-		},
-		MobileDependency{},
-	)
-
-	engine.dnsServer = &dns.MockServer{
-		UpdateDNSServerFunc: func(serial uint64, update nbdns.Config) error { return nil },
-	}
-
-	err = engine.Start(nil, nil)
-	require.NoError(t, err)
-
-	defer func() {
-		err := engine.Stop()
-		if err != nil {
-			return
-		}
-	}()
-
-	peerWithSSH := &mgmtProto.RemotePeerConfig{
-		WgPubKey:   "MNHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
-		AllowedIps: []string{"100.64.0.21/24"},
-		SshConfig: &mgmtProto.SSHConfig{
-			SshPubKey: []byte("ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIFATYCqaQw/9id1Qkq3n16JYhDhXraI6Pc1fgB8ynEfQ"),
-		},
-	}
-
-	// SSH server is not enabled so SSH config of a remote peer should be ignored
-	networkMap := &mgmtProto.NetworkMap{
-		Serial:             6,
-		PeerConfig:         nil,
-		RemotePeers:        []*mgmtProto.RemotePeerConfig{peerWithSSH},
-		RemotePeersIsEmpty: false,
-	}
-
-	_, err = engine.updateNetworkMap(networkMap, maxPeersPerSyncPass, true)
-	require.NoError(t, err)
-
-	assert.Nil(t, engine.sshServer)
-
-	// SSH server is enabled, therefore SSH config should be applied
-	networkMap = &mgmtProto.NetworkMap{
-		Serial: 7,
-		PeerConfig: &mgmtProto.PeerConfig{Address: "100.64.0.1/24",
-			SshConfig: &mgmtProto.SSHConfig{
-				SshEnabled: true,
-				JwtConfig: &mgmtProto.JWTConfig{
-					Issuer:       "test-issuer",
-					Audience:     "test-audience",
-					KeysLocation: "test-keys",
-					MaxTokenAge:  3600,
-				},
-			}},
-		RemotePeers:        []*mgmtProto.RemotePeerConfig{peerWithSSH},
-		RemotePeersIsEmpty: false,
-	}
-
-	_, err = engine.updateNetworkMap(networkMap, maxPeersPerSyncPass, true)
-	require.NoError(t, err)
-
-	time.Sleep(250 * time.Millisecond)
-	assert.NotNil(t, engine.sshServer)
-
-	// now remove peer
-	networkMap = &mgmtProto.NetworkMap{
-		Serial:             8,
-		RemotePeers:        []*mgmtProto.RemotePeerConfig{},
-		RemotePeersIsEmpty: false,
-	}
-
-	_, err = engine.updateNetworkMap(networkMap, maxPeersPerSyncPass, true)
-	require.NoError(t, err)
-
-	// time.Sleep(250 * time.Millisecond)
-	assert.NotNil(t, engine.sshServer)
-
-	// now disable SSH server
-	networkMap = &mgmtProto.NetworkMap{
-		Serial: 9,
-		PeerConfig: &mgmtProto.PeerConfig{Address: "100.64.0.1/24",
-			SshConfig: &mgmtProto.SSHConfig{SshEnabled: false}},
-		RemotePeers:        []*mgmtProto.RemotePeerConfig{peerWithSSH},
-		RemotePeersIsEmpty: false,
-	}
-
-	_, err = engine.updateNetworkMap(networkMap, maxPeersPerSyncPass, true)
-	require.NoError(t, err)
-
-	assert.Nil(t, engine.sshServer)
-}
-
-func TestEngine_Sync(t *testing.T) {
-	key, err := wgtypes.GeneratePrivateKey()
-	if err != nil {
-		t.Fatal(err)
-		return
-	}
-
-	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
-	defer cancel()
-
-	// feed updates to Engine via mocked Management client
-	updates := make(chan *mgmtProto.SyncResponse)
-	defer close(updates)
-	syncFunc := func(ctx context.Context, info *system.Info, msgHandler func(msg *mgmtProto.SyncResponse) error) error {
-		for msg := range updates {
-			err := msgHandler(msg)
-			if err != nil {
-				t.Fatal(err)
-			}
-		}
-		return nil
-	}
-	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
-	engine := NewEngine(ctx, cancel, &EngineConfig{
-		WgIfaceName:  "utun103",
-		WgAddr:       wgaddr.MustParseWGAddress("100.64.0.1/24"),
-		WgPrivateKey: key,
-		WgPort:       33100,
-		MTU:          iface.DefaultMTU,
-	}, EngineServices{
-		SignalClient:   &signal.MockClient{},
-		MgmClient:      &mgmt.MockClient{SyncFunc: syncFunc},
-		RelayManager:   relayMgr,
-		StatusRecorder: peer.NewRecorder("https://mgm"),
-	}, MobileDependency{})
-	engine.ctx = ctx
-
-	engine.dnsServer = &dns.MockServer{
-		UpdateDNSServerFunc: func(serial uint64, update nbdns.Config) error { return nil },
-	}
-
-	defer func() {
-		err := engine.Stop()
-		if err != nil {
-			return
-		}
-	}()
-
-	err = engine.Start(nil, nil)
-	if err != nil {
-		t.Fatal(err)
-		return
-	}
-
-	peer1 := &mgmtProto.RemotePeerConfig{
-		WgPubKey:   "RRHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
-		AllowedIps: []string{"100.64.0.10/24"},
-	}
-	peer2 := &mgmtProto.RemotePeerConfig{
-		WgPubKey:   "LLHf3Ma6z6mdLbriAJbqhX9+nM/B71lgw2+91q3LlhU=",
-		AllowedIps: []string{"100.64.0.11/24"},
-	}
-	peer3 := &mgmtProto.RemotePeerConfig{
-		WgPubKey:   "GGHf3Ma6z6mdLbriAJbqhX9+nM/B71lgw2+91q3LlhU=",
-		AllowedIps: []string{"100.64.0.12/24"},
-	}
-	// 1st update with just 1 peer and serial larger than the current serial of the engine => apply update
-	updates <- &mgmtProto.SyncResponse{
-		NetworkMap: &mgmtProto.NetworkMap{
-			Serial:             10,
-			PeerConfig:         nil,
-			RemotePeers:        []*mgmtProto.RemotePeerConfig{peer1, peer2, peer3},
-			RemotePeersIsEmpty: false,
-		},
-	}
-
-	timeout := time.After(time.Second * 2)
-	for {
-		select {
-		case <-timeout:
-			t.Fatalf("timeout while waiting for test to finish")
-			return
-		default:
-		}
-
-		if getPeers(engine) == 3 && engine.networkSerial == 10 {
-			break
-		}
-	}
-}
-
-func TestEngine_MultiplePeers(t *testing.T) {
-	// log.SetLevel(log.DebugLevel)
-
-	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
-	defer cancel()
-
-	sigServer, signalAddr, err := startSignal(t)
-	if err != nil {
-		t.Fatal(err)
-		return
-	}
-	defer sigServer.Stop()
-	mgmtServer, mgmtAddr, err := startManagement(t, t.TempDir(), "../testdata/store.sql")
-	if err != nil {
-		t.Fatal(err)
-		return
-	}
-	defer mgmtServer.GracefulStop()
-
-	setupKey := "A2C8E62B-38F5-4553-B31E-DD66C696CEBB"
-
-	mu := sync.Mutex{}
-	engines := []*Engine{}
-	numPeers := 10
-	wg := sync.WaitGroup{}
-	wg.Add(numPeers)
-	// create and start peers
-	for i := 0; i < numPeers; i++ {
-		j := i
-		go func() {
-			engine, err := createEngine(ctx, cancel, setupKey, j, mgmtAddr, signalAddr)
-			if err != nil {
-				wg.Done()
-				t.Errorf("unable to create the engine for peer %d with error %v", j, err)
-				return
-			}
-			engine.dnsServer = &dns.MockServer{}
-			mu.Lock()
-			defer mu.Unlock()
-			guid := fmt.Sprintf("{%s}", uuid.New().String())
-			device.CustomWindowsGUIDString = strings.ToLower(guid)
-			err = engine.Start(nil, nil)
-			if err != nil {
-				t.Errorf("unable to start engine for peer %d with error %v", j, err)
-				wg.Done()
-				return
-			}
-			engines = append(engines, engine)
-			wg.Done()
-		}()
-	}
-
-	// wait until all have been created and started
-	wg.Wait()
-	if len(engines) != numPeers {
-		t.Fatal("not all peers were started")
-	}
-	// check whether all the peer have expected peers connected
-
-	expectedConnected := numPeers * (numPeers - 1)
-
-	// adjust according to timeouts
-	timeout := 50 * time.Second
-	timeoutChan := time.After(timeout)
-	ticker := time.NewTicker(time.Second)
-	defer ticker.Stop()
-loop:
-	for {
-		select {
-		case <-timeoutChan:
-			t.Fatalf("waiting for expected connections timeout after %s", timeout.String())
-			break loop
-		case <-ticker.C:
-			totalConnected := 0
-			for _, engine := range engines {
-				totalConnected += getConnectedPeers(engine)
-			}
-			if totalConnected == expectedConnected {
-				log.Infof("total connected=%d", totalConnected)
-				break loop
-			}
-			log.Infof("total connected=%d", totalConnected)
-		}
-	}
-	// cleanup test
-	for n, peerEngine := range engines {
-		t.Logf("stopping peer with interface %s from multipeer test, loopIndex %d", peerEngine.wgInterface.Name(), n)
-		errStop := peerEngine.mgmClient.Close()
-		if errStop != nil {
-			log.Infoln("got error trying to close management clients from engine: ", errStop)
-		}
-		errStop = peerEngine.Stop()
-		if errStop != nil {
-			log.Infoln("got error trying to close testing peers engine: ", errStop)
-		}
-	}
-}
-
-var (
-	kaep = keepalive.EnforcementPolicy{
-		MinTime:             15 * time.Second,
-		PermitWithoutStream: true,
-	}
-
-	kasp = keepalive.ServerParameters{
-		MaxConnectionIdle:     15 * time.Second,
-		MaxConnectionAgeGrace: 5 * time.Second,
-		Time:                  5 * time.Second,
-		Timeout:               2 * time.Second,
-	}
-)
-
-func createEngine(ctx context.Context, cancel context.CancelFunc, setupKey string, i int, mgmtAddr string, signalAddr string) (*Engine, error) {
-	key, err := wgtypes.GeneratePrivateKey()
-	if err != nil {
-		return nil, err
-	}
-	mgmtClient, err := mgmt.NewClient(ctx, mgmtAddr, key, false)
-	if err != nil {
-		return nil, err
-	}
-	signalClient, err := signal.NewClient(ctx, signalAddr, key, false)
-	if err != nil {
-		return nil, err
-	}
-
-	info := system.GetInfo(ctx)
-	resp, err := mgmtClient.Register(setupKey, "", info, nil, nil)
-	if err != nil {
-		return nil, err
-	}
-
-	var ifaceName string
-	if runtime.GOOS == "darwin" {
-		ifaceName = fmt.Sprintf("utun1%d", i)
-	} else {
-		ifaceName = fmt.Sprintf("wt%d", i)
-	}
-
-	wgPort := 33100 + i
-	conf := &EngineConfig{
-		WgIfaceName:  ifaceName,
-		WgAddr:       wgaddr.MustParseWGAddress(resp.PeerConfig.Address),
-		WgPrivateKey: key,
-		WgPort:       wgPort,
-		MTU:          iface.DefaultMTU,
-	}
-
-	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
-	e, err := NewEngine(ctx, cancel, conf, EngineServices{
-		SignalClient:   signalClient,
-		MgmClient:      mgmtClient,
-		RelayManager:   relayMgr,
-		StatusRecorder: peer.NewRecorder("https://mgm"),
-	}, MobileDependency{}), nil
-	e.ctx = ctx
-	return e, err
-}
-
-func startSignal(t *testing.T) (*grpc.Server, string, error) {
-	t.Helper()
-
-	s := grpc.NewServer(grpc.KeepaliveEnforcementPolicy(kaep), grpc.KeepaliveParams(kasp))
-
-	lis, err := net.Listen("tcp", "localhost:0")
-	if err != nil {
-		log.Fatalf("failed to listen: %v", err)
-	}
-
-	srv, err := signalServer.NewServer(context.Background(), otel.Meter(""))
-	require.NoError(t, err)
-	proto.RegisterSignalExchangeServer(s, srv)
-
-	go func() {
-		if err = s.Serve(lis); err != nil {
-			log.Fatalf("failed to serve: %v", err)
-		}
-	}()
-
-	return s, lis.Addr().String(), nil
-}
-
-func startManagement(t *testing.T, dataDir, testFile string) (*grpc.Server, string, error) {
-	t.Helper()
-
-	config := &config.Config{
-		Stuns:      []*config.Host{},
-		TURNConfig: &config.TURNConfig{},
-		Relay: &config.Relay{
-			Addresses:      []string{"127.0.0.1:1234"},
-			CredentialsTTL: util.Duration{Duration: time.Hour},
-			Secret:         "222222222222222222",
-		},
-		Signal: &config.Host{
-			Proto: "http",
-			URI:   "localhost:10000",
-		},
-		Datadir:    dataDir,
-		HttpConfig: nil,
-	}
-
-	lis, err := net.Listen("tcp", "localhost:0")
-	if err != nil {
-		return nil, "", err
-	}
-	s := grpc.NewServer(grpc.KeepaliveEnforcementPolicy(kaep), grpc.KeepaliveParams(kasp))
-
-	store, cleanUp, err := store.NewTestStoreFromSQL(context.Background(), testFile, config.Datadir)
-	if err != nil {
-		return nil, "", err
-	}
-	t.Cleanup(cleanUp)
-
-	eventStore := &activity.InMemoryEventStore{}
-	if err != nil {
-		return nil, "", err
-	}
-
-	permissionsManager := permissions.NewManager(store)
-	peersManager := peers.NewManager(store, permissionsManager)
-	jobManager := job.NewJobManager(nil, store, peersManager)
-
-	cacheStore, err := nbcache.NewStore(context.Background(), 100*time.Millisecond, 300*time.Millisecond, 100)
-	if err != nil {
-		return nil, "", err
-	}
-
-	ia, _ := validator.NewIntegratedValidator(context.Background(), peersManager, nil, eventStore, cacheStore)
-
-	metrics, err := telemetry.NewDefaultAppMetrics(context.Background())
-	require.NoError(t, err)
-
-	ctrl := gomock.NewController(t)
-	t.Cleanup(ctrl.Finish)
-	settingsMockManager := settings.NewMockManager(ctrl)
-	settingsMockManager.EXPECT().
-		GetSettings(gomock.Any(), gomock.Any(), gomock.Any()).
-		Return(&types.Settings{}, nil).
-		AnyTimes()
-	settingsMockManager.EXPECT().
-		GetExtraSettings(gomock.Any(), gomock.Any()).
-		Return(&types.ExtraSettings{}, nil).
-		AnyTimes()
-
-	groupsManager := groups.NewManagerMock()
-
-	updateManager := update_channel.NewPeersUpdateManager(metrics)
-	requestBuffer := server.NewAccountRequestBuffer(context.Background(), store)
-	networkMapController := controller.NewController(context.Background(), store, metrics, updateManager, requestBuffer, server.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersManager), config)
-	accountManager, err := server.BuildManager(context.Background(), config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, ia, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManager, false, cacheStore)
-	if err != nil {
-		return nil, "", err
-	}
-
-	secretsManager, err := nbgrpc.NewTimeBasedAuthSecretsManager(updateManager, config.TURNConfig, config.Relay, settingsMockManager, groupsManager)
-	if err != nil {
-		return nil, "", err
-	}
-	mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, &server.MockIntegratedValidator{}, networkMapController, nil, nil)
-	if err != nil {
-		return nil, "", err
-	}
-	mgmtProto.RegisterManagementServiceServer(s, mgmtServer)
-	go func() {
-		if err = s.Serve(lis); err != nil {
-			log.Fatalf("failed to serve: %v", err)
-		}
-	}()
-
-	return s, lis.Addr().String(), nil
-}
-
-// getConnectedPeers returns a connection Status or nil if peer connection wasn't found
-func getConnectedPeers(e *Engine) int {
-	e.syncMsgMux.Lock()
-	defer e.syncMsgMux.Unlock()
-	i := 0
-	for _, id := range e.peerStore.PeersPubKey() {
-		conn, _ := e.peerStore.PeerConn(id)
-		if conn.IsConnected() {
-			i++
-		}
-	}
-	return i
-}
-
-func getPeers(e *Engine) int {
-	e.syncMsgMux.Lock()
-	defer e.syncMsgMux.Unlock()
-
-	return len(e.peerStore.PeersPubKey())
-}

client/internal/engine_session_deadline_test.go

@@ -0,0 +1,78 @@
+package internal
+
+import (
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+	"google.golang.org/protobuf/types/known/timestamppb"
+
+	"github.com/netbirdio/netbird/client/internal/auth/sessionwatch"
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+// TestApplySessionDeadline_ThreeState pins down the 3-state semantics of the
+// wire field carried on LoginResponse / SyncResponse:
+//
+//   - nil pointer            → no info; previously-anchored deadline survives
+//   - explicit zero value    → "expiry disabled" sentinel; both sinks cleared
+//   - valid future timestamp → new deadline propagated to both sinks
+func TestApplySessionDeadline_ThreeState(t *testing.T) {
+	newEngine := func() *Engine {
+		recorder := peer.NewRecorder("")
+		return &Engine{
+			statusRecorder: recorder,
+			sessionWatcher: sessionwatch.New(recorder),
+		}
+	}
+
+	t.Run("valid timestamp sets deadline on both sinks", func(t *testing.T) {
+		e := newEngine()
+		deadline := time.Now().Add(time.Hour).UTC().Truncate(time.Second)
+
+		e.ApplySessionDeadline(timestamppb.New(deadline))
+
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().Equal(deadline),
+			"status recorder should hold the new deadline")
+	})
+
+	t.Run("nil is a no-op and preserves previous deadline", func(t *testing.T) {
+		e := newEngine()
+		seeded := time.Now().Add(time.Hour).UTC().Truncate(time.Second)
+		e.ApplySessionDeadline(timestamppb.New(seeded))
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().Equal(seeded))
+
+		e.ApplySessionDeadline(nil)
+
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().Equal(seeded),
+			"nil snapshot must not disturb the existing deadline")
+	})
+
+	t.Run("explicit zero clears a previously-anchored deadline", func(t *testing.T) {
+		e := newEngine()
+		seeded := time.Now().Add(time.Hour).UTC().Truncate(time.Second)
+		e.ApplySessionDeadline(timestamppb.New(seeded))
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().Equal(seeded))
+
+		// Explicit zero Timestamp{} (seconds=0, nanos=0) is the
+		// "expiry disabled / not SSO" sentinel.
+		e.ApplySessionDeadline(&timestamppb.Timestamp{})
+
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().IsZero(),
+			"explicit zero sentinel must clear the deadline")
+	})
+
+	t.Run("invalid timestamp clears the deadline", func(t *testing.T) {
+		e := newEngine()
+		seeded := time.Now().Add(time.Hour).UTC().Truncate(time.Second)
+		e.ApplySessionDeadline(timestamppb.New(seeded))
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().Equal(seeded))
+
+		// Out-of-range nanos → IsValid()==false; same-meaning as the
+		// disabled sentinel for downstream sinks.
+		e.ApplySessionDeadline(&timestamppb.Timestamp{Seconds: 1, Nanos: -1})
+
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().IsZero(),
+			"invalid timestamp must clear the deadline")
+	})
+}

client/internal/engine_sessionwatch.go

@@ -0,0 +1,16 @@
+//go:build !js
+
+package internal
+
+import (
+	"github.com/netbirdio/netbird/client/internal/auth/sessionwatch"
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+// newSessionWatcher returns the real SSO session expiry watcher for every
+// non-wasm build. The js/wasm build gets a no-op stub from
+// engine_sessionwatch_js.go so the sessionwatch package (and its timer
+// machinery) never links into the wasm binary.
+func newSessionWatcher(recorder *peer.Status) sessionDeadlineWatcher {
+	return sessionwatch.New(recorder)
+}

client/internal/engine_sessionwatch_js.go

@@ -0,0 +1,44 @@
+//go:build js
+
+package internal
+
+import (
+	"time"
+
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+// noopSessionWatcher is the js/wasm stand-in for sessionwatch.Watcher. The
+// wasm client never runs the engine's session-warning flow (the interactive
+// T-WarningLead notification and the T-FinalWarningLead fallback dialog live
+// in the desktop UI), so linking the full sessionwatch package (timers, event
+// composition) would only bloat the binary.
+//
+// It still mirrors the deadline into the status recorder so the SubscribeStatus
+// / Status snapshot the UI consumes stays correct — only the timer-driven
+// warnings are dropped.
+type noopSessionWatcher struct {
+	recorder *peer.Status
+}
+
+func newSessionWatcher(recorder *peer.Status) sessionDeadlineWatcher {
+	return noopSessionWatcher{recorder: recorder}
+}
+
+// Update mirrors the real watcher's recorder propagation without the timers or
+// sanity-check sentinels: a valid deadline is exposed on the status snapshot,
+// the zero time clears it.
+func (w noopSessionWatcher) Update(deadline time.Time) error {
+	if w.recorder != nil {
+		w.recorder.SetSessionExpiresAt(deadline)
+	}
+	return nil
+}
+
+func (noopSessionWatcher) Dismiss() {
+	// No-op: only suppresses the timer-driven final-warning, which this stub never arms.
+}
+
+func (noopSessionWatcher) Close() {
+	// No-op: no timers to stop and no state to unwind; the recorder is cleared via Update(zero).
+}

client/internal/engine_test.go

@@ -6,18 +6,37 @@ import (
 	"net"
 	"net/netip"
 	"os"
+	"runtime"
 	"strings"
 	"sync"
 	"testing"
 	"time"
 
+	"github.com/golang/mock/gomock"
+	"github.com/google/uuid"
+	log "github.com/sirupsen/logrus"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
+	"go.opentelemetry.io/otel"
 	wgdevice "golang.zx2c4.com/wireguard/device"
 	"golang.zx2c4.com/wireguard/tun/netstack"
 	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/keepalive"
 
 	"github.com/netbirdio/netbird/client/internal/stdnet"
+	"github.com/netbirdio/netbird/management/server/job"
+
+	"github.com/netbirdio/netbird/management/server/integrations/integrated_validator/validator"
+
+	"github.com/netbirdio/netbird/management/internals/controllers/network_map/controller"
+	"github.com/netbirdio/netbird/management/internals/controllers/network_map/update_channel"
+	"github.com/netbirdio/netbird/management/internals/modules/peers"
+	"github.com/netbirdio/netbird/management/internals/modules/peers/ephemeral/manager"
+	nbgrpc "github.com/netbirdio/netbird/management/internals/shared/grpc"
+
+	"github.com/netbirdio/netbird/management/internals/server/config"
+	"github.com/netbirdio/netbird/management/server/groups"
 
 	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/iface/configurer"
@@ -31,7 +50,18 @@ import (
 	icemaker "github.com/netbirdio/netbird/client/internal/peer/ice"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/internal/routemanager"
+	nbssh "github.com/netbirdio/netbird/client/ssh"
+	"github.com/netbirdio/netbird/client/system"
 	nbdns "github.com/netbirdio/netbird/dns"
+	"github.com/netbirdio/netbird/management/server"
+	"github.com/netbirdio/netbird/management/server/activity"
+	nbcache "github.com/netbirdio/netbird/management/server/cache"
+	"github.com/netbirdio/netbird/management/server/integrations/port_forwarding"
+	"github.com/netbirdio/netbird/management/server/permissions"
+	"github.com/netbirdio/netbird/management/server/settings"
+	"github.com/netbirdio/netbird/management/server/store"
+	"github.com/netbirdio/netbird/management/server/telemetry"
+	"github.com/netbirdio/netbird/management/server/types"
 	"github.com/netbirdio/netbird/monotime"
 	"github.com/netbirdio/netbird/route"
 	mgmt "github.com/netbirdio/netbird/shared/management/client"
@@ -39,9 +69,25 @@ import (
 	"github.com/netbirdio/netbird/shared/netiputil"
 	relayClient "github.com/netbirdio/netbird/shared/relay/client"
 	signal "github.com/netbirdio/netbird/shared/signal/client"
+	"github.com/netbirdio/netbird/shared/signal/proto"
+	signalServer "github.com/netbirdio/netbird/signal/server"
 	"github.com/netbirdio/netbird/util"
 )
 
+var (
+	kaep = keepalive.EnforcementPolicy{
+		MinTime:             15 * time.Second,
+		PermitWithoutStream: true,
+	}
+
+	kasp = keepalive.ServerParameters{
+		MaxConnectionIdle:     15 * time.Second,
+		MaxConnectionAgeGrace: 5 * time.Second,
+		Time:                  5 * time.Second,
+		Timeout:               2 * time.Second,
+	}
+)
+
 type MockWGIface struct {
 	CreateFunc                 func() error
 	CreateOnAndroidFunc        func(routeRange []string, ip string, domains []string) error
@@ -178,10 +224,6 @@ func (m *MockWGIface) LastActivities() map[string]monotime.Time {
 	return nil
 }
 
-func (m *MockWGIface) MTU() uint16 {
-	return 1280
-}
-
 func (m *MockWGIface) SetPresharedKey(peerKey string, psk wgtypes.Key, updateOnly bool) error {
 	return nil
 }
@@ -192,6 +234,129 @@ func TestMain(m *testing.M) {
 	os.Exit(code)
 }
 
+func TestEngine_SSH(t *testing.T) {
+	key, err := wgtypes.GeneratePrivateKey()
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+
+	sshKey, err := nbssh.GeneratePrivateKey(nbssh.ED25519)
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
+	engine := NewEngine(
+		ctx, cancel,
+		&EngineConfig{
+			WgIfaceName:      "utun101",
+			WgAddr:           wgaddr.MustParseWGAddress("100.64.0.1/24"),
+			WgPrivateKey:     key,
+			WgPort:           33100,
+			ServerSSHAllowed: true,
+			MTU:              iface.DefaultMTU,
+			SSHKey:           sshKey,
+		},
+		EngineServices{
+			SignalClient:   &signal.MockClient{},
+			MgmClient:      &mgmt.MockClient{},
+			RelayManager:   relayMgr,
+			StatusRecorder: peer.NewRecorder("https://mgm"),
+		},
+		MobileDependency{},
+	)
+
+	engine.dnsServer = &dns.MockServer{
+		UpdateDNSServerFunc: func(serial uint64, update nbdns.Config) error { return nil },
+	}
+
+	err = engine.Start(nil, nil)
+	require.NoError(t, err)
+
+	defer func() {
+		err := engine.Stop()
+		if err != nil {
+			return
+		}
+	}()
+
+	peerWithSSH := &mgmtProto.RemotePeerConfig{
+		WgPubKey:   "MNHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
+		AllowedIps: []string{"100.64.0.21/24"},
+		SshConfig: &mgmtProto.SSHConfig{
+			SshPubKey: []byte("ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIFATYCqaQw/9id1Qkq3n16JYhDhXraI6Pc1fgB8ynEfQ"),
+		},
+	}
+
+	// SSH server is not enabled so SSH config of a remote peer should be ignored
+	networkMap := &mgmtProto.NetworkMap{
+		Serial:             6,
+		PeerConfig:         nil,
+		RemotePeers:        []*mgmtProto.RemotePeerConfig{peerWithSSH},
+		RemotePeersIsEmpty: false,
+	}
+
+	err = engine.updateNetworkMap(networkMap)
+	require.NoError(t, err)
+
+	assert.Nil(t, engine.sshServer)
+
+	// SSH server is enabled, therefore SSH config should be applied
+	networkMap = &mgmtProto.NetworkMap{
+		Serial: 7,
+		PeerConfig: &mgmtProto.PeerConfig{Address: "100.64.0.1/24",
+			SshConfig: &mgmtProto.SSHConfig{
+				SshEnabled: true,
+				JwtConfig: &mgmtProto.JWTConfig{
+					Issuer:       "test-issuer",
+					Audience:     "test-audience",
+					KeysLocation: "test-keys",
+					MaxTokenAge:  3600,
+				},
+			}},
+		RemotePeers:        []*mgmtProto.RemotePeerConfig{peerWithSSH},
+		RemotePeersIsEmpty: false,
+	}
+
+	err = engine.updateNetworkMap(networkMap)
+	require.NoError(t, err)
+
+	time.Sleep(250 * time.Millisecond)
+	assert.NotNil(t, engine.sshServer)
+
+	// now remove peer
+	networkMap = &mgmtProto.NetworkMap{
+		Serial:             8,
+		RemotePeers:        []*mgmtProto.RemotePeerConfig{},
+		RemotePeersIsEmpty: false,
+	}
+
+	err = engine.updateNetworkMap(networkMap)
+	require.NoError(t, err)
+
+	// time.Sleep(250 * time.Millisecond)
+	assert.NotNil(t, engine.sshServer)
+
+	// now disable SSH server
+	networkMap = &mgmtProto.NetworkMap{
+		Serial: 9,
+		PeerConfig: &mgmtProto.PeerConfig{Address: "100.64.0.1/24",
+			SshConfig: &mgmtProto.SSHConfig{SshEnabled: false}},
+		RemotePeers:        []*mgmtProto.RemotePeerConfig{peerWithSSH},
+		RemotePeersIsEmpty: false,
+	}
+
+	err = engine.updateNetworkMap(networkMap)
+	require.NoError(t, err)
+
+	assert.Nil(t, engine.sshServer)
+}
+
 func TestEngine_SSHUpdateLogic(t *testing.T) {
 	// Test that SSH server start/stop logic works based on config
 	engine := &Engine{
@@ -261,7 +426,7 @@ func TestEngine_UpdateNetworkMap(t *testing.T) {
 		return
 	}
 
-	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
+	ctx, cancel := context.WithCancel(context.Background())
 	defer cancel()
 
 	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
@@ -437,7 +602,7 @@ func TestEngine_UpdateNetworkMap(t *testing.T) {
 
 	for _, c := range []testCase{case1, case2, case3, case4, case5, case6} {
 		t.Run(c.name, func(t *testing.T) {
-			_, err = engine.updateNetworkMap(c.networkMap, maxPeersPerSyncPass, true)
+			err = engine.updateNetworkMap(c.networkMap)
 			if err != nil {
 				t.Fatal(err)
 				return
@@ -464,47 +629,97 @@ func TestEngine_UpdateNetworkMap(t *testing.T) {
 			}
 		})
 	}
+}
 
-	// chunked apply: with a per-pass cap smaller than the number of peers, a
-	// single updateNetworkMap applies one batch and reports more==true; the
-	// caller re-runs until convergence. (engine currently holds 0 peers.)
-	t.Run("chunked add converges over multiple passes", func(t *testing.T) {
-		nm := &mgmtProto.NetworkMap{
-			Serial:      6,
-			RemotePeers: []*mgmtProto.RemotePeerConfig{peer1, peer2, peer3},
+func TestEngine_Sync(t *testing.T) {
+	key, err := wgtypes.GeneratePrivateKey()
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// feed updates to Engine via mocked Management client
+	updates := make(chan *mgmtProto.SyncResponse)
+	defer close(updates)
+	syncFunc := func(ctx context.Context, info *system.Info, msgHandler func(msg *mgmtProto.SyncResponse) error) error {
+		for msg := range updates {
+			err := msgHandler(msg)
+			if err != nil {
+				t.Fatal(err)
+			}
+		}
+		return nil
+	}
+	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
+	engine := NewEngine(ctx, cancel, &EngineConfig{
+		WgIfaceName:  "utun103",
+		WgAddr:       wgaddr.MustParseWGAddress("100.64.0.1/24"),
+		WgPrivateKey: key,
+		WgPort:       33100,
+		MTU:          iface.DefaultMTU,
+	}, EngineServices{
+		SignalClient:   &signal.MockClient{},
+		MgmClient:      &mgmt.MockClient{SyncFunc: syncFunc},
+		RelayManager:   relayMgr,
+		StatusRecorder: peer.NewRecorder("https://mgm"),
+	}, MobileDependency{})
+	engine.ctx = ctx
+
+	engine.dnsServer = &dns.MockServer{
+		UpdateDNSServerFunc: func(serial uint64, update nbdns.Config) error { return nil },
+	}
+
+	defer func() {
+		err := engine.Stop()
+		if err != nil {
+			return
+		}
+	}()
+
+	err = engine.Start(nil, nil)
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+
+	peer1 := &mgmtProto.RemotePeerConfig{
+		WgPubKey:   "RRHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
+		AllowedIps: []string{"100.64.0.10/24"},
+	}
+	peer2 := &mgmtProto.RemotePeerConfig{
+		WgPubKey:   "LLHf3Ma6z6mdLbriAJbqhX9+nM/B71lgw2+91q3LlhU=",
+		AllowedIps: []string{"100.64.0.11/24"},
+	}
+	peer3 := &mgmtProto.RemotePeerConfig{
+		WgPubKey:   "GGHf3Ma6z6mdLbriAJbqhX9+nM/B71lgw2+91q3LlhU=",
+		AllowedIps: []string{"100.64.0.12/24"},
+	}
+	// 1st update with just 1 peer and serial larger than the current serial of the engine => apply update
+	updates <- &mgmtProto.SyncResponse{
+		NetworkMap: &mgmtProto.NetworkMap{
+			Serial:             10,
+			PeerConfig:         nil,
+			RemotePeers:        []*mgmtProto.RemotePeerConfig{peer1, peer2, peer3},
+			RemotePeersIsEmpty: false,
+		},
+	}
+
+	timeout := time.After(time.Second * 2)
+	for {
+		select {
+		case <-timeout:
+			t.Fatalf("timeout while waiting for test to finish")
+			return
+		default:
 		}
 
-		more, err := engine.updateNetworkMap(nm, 1, true)
-		require.NoError(t, err)
-		require.True(t, more, "pass 1 should signal more")
-		require.Len(t, engine.peerStore.PeersPubKey(), 1)
-
-		more, err = engine.updateNetworkMap(nm, 1, false)
-		require.NoError(t, err)
-		require.True(t, more, "pass 2 should signal more")
-		require.Len(t, engine.peerStore.PeersPubKey(), 2)
-
-		more, err = engine.updateNetworkMap(nm, 1, false)
-		require.NoError(t, err)
-		require.False(t, more, "pass 3 should converge")
-		require.Len(t, engine.peerStore.PeersPubKey(), 3)
-	})
-
-	t.Run("chunked remove converges over multiple passes", func(t *testing.T) {
-		nm := &mgmtProto.NetworkMap{
-			Serial:      7,
-			RemotePeers: []*mgmtProto.RemotePeerConfig{peer1}, // remove peer2, peer3
+		if getPeers(engine) == 3 && engine.networkSerial == 10 {
+			break
 		}
-
-		more, err := engine.updateNetworkMap(nm, 1, true)
-		require.NoError(t, err)
-		require.True(t, more, "pass 1 should signal more (2 to remove, cap 1)")
-
-		more, err = engine.updateNetworkMap(nm, 1, false)
-		require.NoError(t, err)
-		require.False(t, more, "pass 2 should converge")
-		require.Len(t, engine.peerStore.PeersPubKey(), 1)
-	})
+	}
 }
 
 func TestEngine_UpdateNetworkMapWithRoutes(t *testing.T) {
@@ -602,7 +817,7 @@ func TestEngine_UpdateNetworkMapWithRoutes(t *testing.T) {
 				return
 			}
 
-			ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
+			ctx, cancel := context.WithCancel(context.Background())
 			defer cancel()
 
 			wgIfaceName := fmt.Sprintf("utun%d", 104+n)
@@ -675,7 +890,7 @@ func TestEngine_UpdateNetworkMapWithRoutes(t *testing.T) {
 				}
 			}()
 
-			_, err = engine.updateNetworkMap(testCase.networkMap, maxPeersPerSyncPass, true)
+			err = engine.updateNetworkMap(testCase.networkMap)
 			assert.NoError(t, err, "shouldn't return error")
 			assert.Equal(t, testCase.expectedSerial, input.inputSerial, "serial should match")
 			assert.Len(t, input.clientRoutes, testCase.expectedLen, "clientRoutes len should match")
@@ -809,7 +1024,7 @@ func TestEngine_UpdateNetworkMapWithDNSUpdate(t *testing.T) {
 				return
 			}
 
-			ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
+			ctx, cancel := context.WithCancel(context.Background())
 			defer cancel()
 
 			wgIfaceName := fmt.Sprintf("utun%d", 104+n)
@@ -879,7 +1094,7 @@ func TestEngine_UpdateNetworkMapWithDNSUpdate(t *testing.T) {
 				}
 			}()
 
-			_, err = engine.updateNetworkMap(testCase.networkMap, maxPeersPerSyncPass, true)
+			err = engine.updateNetworkMap(testCase.networkMap)
 			assert.NoError(t, err, "shouldn't return error")
 			assert.Equal(t, testCase.expectedSerial, input.inputSerial, "serial should match")
 			assert.Len(t, input.inputNSGroups, testCase.expectedZonesLen, "zones len should match")
@@ -890,6 +1105,104 @@ func TestEngine_UpdateNetworkMapWithDNSUpdate(t *testing.T) {
 	}
 }
 
+func TestEngine_MultiplePeers(t *testing.T) {
+	// log.SetLevel(log.DebugLevel)
+
+	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
+	defer cancel()
+
+	sigServer, signalAddr, err := startSignal(t)
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+	defer sigServer.Stop()
+	mgmtServer, mgmtAddr, err := startManagement(t, t.TempDir(), "../testdata/store.sql")
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+	defer mgmtServer.GracefulStop()
+
+	setupKey := "A2C8E62B-38F5-4553-B31E-DD66C696CEBB"
+
+	mu := sync.Mutex{}
+	engines := []*Engine{}
+	numPeers := 10
+	wg := sync.WaitGroup{}
+	wg.Add(numPeers)
+	// create and start peers
+	for i := 0; i < numPeers; i++ {
+		j := i
+		go func() {
+			engine, err := createEngine(ctx, cancel, setupKey, j, mgmtAddr, signalAddr)
+			if err != nil {
+				wg.Done()
+				t.Errorf("unable to create the engine for peer %d with error %v", j, err)
+				return
+			}
+			engine.dnsServer = &dns.MockServer{}
+			mu.Lock()
+			defer mu.Unlock()
+			guid := fmt.Sprintf("{%s}", uuid.New().String())
+			device.CustomWindowsGUIDString = strings.ToLower(guid)
+			err = engine.Start(nil, nil)
+			if err != nil {
+				t.Errorf("unable to start engine for peer %d with error %v", j, err)
+				wg.Done()
+				return
+			}
+			engines = append(engines, engine)
+			wg.Done()
+		}()
+	}
+
+	// wait until all have been created and started
+	wg.Wait()
+	if len(engines) != numPeers {
+		t.Fatal("not all peers was started")
+	}
+	// check whether all the peer have expected peers connected
+
+	expectedConnected := numPeers * (numPeers - 1)
+
+	// adjust according to timeouts
+	timeout := 50 * time.Second
+	timeoutChan := time.After(timeout)
+	ticker := time.NewTicker(time.Second)
+	defer ticker.Stop()
+loop:
+	for {
+		select {
+		case <-timeoutChan:
+			t.Fatalf("waiting for expected connections timeout after %s", timeout.String())
+			break loop
+		case <-ticker.C:
+			totalConnected := 0
+			for _, engine := range engines {
+				totalConnected += getConnectedPeers(engine)
+			}
+			if totalConnected == expectedConnected {
+				log.Infof("total connected=%d", totalConnected)
+				break loop
+			}
+			log.Infof("total connected=%d", totalConnected)
+		}
+	}
+	// cleanup test
+	for n, peerEngine := range engines {
+		t.Logf("stopping peer with interface %s from multipeer test, loopIndex %d", peerEngine.wgInterface.Name(), n)
+		errStop := peerEngine.mgmClient.Close()
+		if errStop != nil {
+			log.Infoln("got error trying to close management clients from engine: ", errStop)
+		}
+		errStop = peerEngine.Stop()
+		if errStop != nil {
+			log.Infoln("got error trying to close testing peers engine: ", errStop)
+		}
+	}
+}
+
 func Test_ParseNATExternalIPMappings(t *testing.T) {
 	ifaceList, err := net.Interfaces()
 	if err != nil {
@@ -1213,6 +1526,187 @@ func TestCompareNetIPLists(t *testing.T) {
 	}
 }
 
+func createEngine(ctx context.Context, cancel context.CancelFunc, setupKey string, i int, mgmtAddr string, signalAddr string) (*Engine, error) {
+	key, err := wgtypes.GeneratePrivateKey()
+	if err != nil {
+		return nil, err
+	}
+	mgmtClient, err := mgmt.NewClient(ctx, mgmtAddr, key, false)
+	if err != nil {
+		return nil, err
+	}
+	signalClient, err := signal.NewClient(ctx, signalAddr, key, false)
+	if err != nil {
+		return nil, err
+	}
+
+	info := system.GetInfo(ctx)
+	resp, err := mgmtClient.Register(setupKey, "", info, nil, nil)
+	if err != nil {
+		return nil, err
+	}
+
+	var ifaceName string
+	if runtime.GOOS == "darwin" {
+		ifaceName = fmt.Sprintf("utun1%d", i)
+	} else {
+		ifaceName = fmt.Sprintf("wt%d", i)
+	}
+
+	wgPort := 33100 + i
+	conf := &EngineConfig{
+		WgIfaceName:  ifaceName,
+		WgAddr:       wgaddr.MustParseWGAddress(resp.PeerConfig.Address),
+		WgPrivateKey: key,
+		WgPort:       wgPort,
+		MTU:          iface.DefaultMTU,
+	}
+
+	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
+	e, err := NewEngine(ctx, cancel, conf, EngineServices{
+		SignalClient:   signalClient,
+		MgmClient:      mgmtClient,
+		RelayManager:   relayMgr,
+		StatusRecorder: peer.NewRecorder("https://mgm"),
+	}, MobileDependency{}), nil
+	e.ctx = ctx
+	return e, err
+}
+
+func startSignal(t *testing.T) (*grpc.Server, string, error) {
+	t.Helper()
+
+	s := grpc.NewServer(grpc.KeepaliveEnforcementPolicy(kaep), grpc.KeepaliveParams(kasp))
+
+	lis, err := net.Listen("tcp", "localhost:0")
+	if err != nil {
+		log.Fatalf("failed to listen: %v", err)
+	}
+
+	srv, err := signalServer.NewServer(context.Background(), otel.Meter(""))
+	require.NoError(t, err)
+	proto.RegisterSignalExchangeServer(s, srv)
+
+	go func() {
+		if err = s.Serve(lis); err != nil {
+			log.Fatalf("failed to serve: %v", err)
+		}
+	}()
+
+	return s, lis.Addr().String(), nil
+}
+
+func startManagement(t *testing.T, dataDir, testFile string) (*grpc.Server, string, error) {
+	t.Helper()
+
+	config := &config.Config{
+		Stuns:      []*config.Host{},
+		TURNConfig: &config.TURNConfig{},
+		Relay: &config.Relay{
+			Addresses:      []string{"127.0.0.1:1234"},
+			CredentialsTTL: util.Duration{Duration: time.Hour},
+			Secret:         "222222222222222222",
+		},
+		Signal: &config.Host{
+			Proto: "http",
+			URI:   "localhost:10000",
+		},
+		Datadir:    dataDir,
+		HttpConfig: nil,
+	}
+
+	lis, err := net.Listen("tcp", "localhost:0")
+	if err != nil {
+		return nil, "", err
+	}
+	s := grpc.NewServer(grpc.KeepaliveEnforcementPolicy(kaep), grpc.KeepaliveParams(kasp))
+
+	store, cleanUp, err := store.NewTestStoreFromSQL(context.Background(), testFile, config.Datadir)
+	if err != nil {
+		return nil, "", err
+	}
+	t.Cleanup(cleanUp)
+
+	eventStore := &activity.InMemoryEventStore{}
+	if err != nil {
+		return nil, "", err
+	}
+
+	permissionsManager := permissions.NewManager(store)
+	peersManager := peers.NewManager(store, permissionsManager)
+	jobManager := job.NewJobManager(nil, store, peersManager)
+
+	cacheStore, err := nbcache.NewStore(context.Background(), 100*time.Millisecond, 300*time.Millisecond, 100)
+	if err != nil {
+		return nil, "", err
+	}
+
+	ia, _ := validator.NewIntegratedValidator(context.Background(), peersManager, nil, eventStore, cacheStore)
+
+	metrics, err := telemetry.NewDefaultAppMetrics(context.Background())
+	require.NoError(t, err)
+
+	ctrl := gomock.NewController(t)
+	t.Cleanup(ctrl.Finish)
+	settingsMockManager := settings.NewMockManager(ctrl)
+	settingsMockManager.EXPECT().
+		GetSettings(gomock.Any(), gomock.Any(), gomock.Any()).
+		Return(&types.Settings{}, nil).
+		AnyTimes()
+	settingsMockManager.EXPECT().
+		GetExtraSettings(gomock.Any(), gomock.Any()).
+		Return(&types.ExtraSettings{}, nil).
+		AnyTimes()
+
+	groupsManager := groups.NewManagerMock()
+
+	updateManager := update_channel.NewPeersUpdateManager(metrics)
+	requestBuffer := server.NewAccountRequestBuffer(context.Background(), store)
+	networkMapController := controller.NewController(context.Background(), store, metrics, updateManager, requestBuffer, server.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersManager), config)
+	accountManager, err := server.BuildManager(context.Background(), config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, ia, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManager, false, cacheStore)
+	if err != nil {
+		return nil, "", err
+	}
+
+	secretsManager, err := nbgrpc.NewTimeBasedAuthSecretsManager(updateManager, config.TURNConfig, config.Relay, settingsMockManager, groupsManager)
+	if err != nil {
+		return nil, "", err
+	}
+	mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, &server.MockIntegratedValidator{}, networkMapController, nil, nil)
+	if err != nil {
+		return nil, "", err
+	}
+	mgmtProto.RegisterManagementServiceServer(s, mgmtServer)
+	go func() {
+		if err = s.Serve(lis); err != nil {
+			log.Fatalf("failed to serve: %v", err)
+		}
+	}()
+
+	return s, lis.Addr().String(), nil
+}
+
+// getConnectedPeers returns a connection Status or nil if peer connection wasn't found
+func getConnectedPeers(e *Engine) int {
+	e.syncMsgMux.Lock()
+	defer e.syncMsgMux.Unlock()
+	i := 0
+	for _, id := range e.peerStore.PeersPubKey() {
+		conn, _ := e.peerStore.PeerConn(id)
+		if conn.IsConnected() {
+			i++
+		}
+	}
+	return i
+}
+
+func getPeers(e *Engine) int {
+	e.syncMsgMux.Lock()
+	defer e.syncMsgMux.Unlock()
+
+	return len(e.peerStore.PeersPubKey())
+}
+
 func mustEncodePrefix(t *testing.T, p netip.Prefix) []byte {
 	t.Helper()
 	b, err := netiputil.EncodePrefix(p)

client/internal/iface_common.go

@@ -44,5 +44,4 @@ type wgIfaceBase interface {
 	FullStats() (*configurer.Stats, error)
 	LastActivities() map[string]monotime.Time
 	SetPresharedKey(peerKey string, psk wgtypes.Key, updateOnly bool) error
-	MTU() uint16
 }

client/internal/lazyconn/activity/listener_bind.go

@@ -119,16 +119,15 @@ func (d *BindListener) ReadPackets() {
 	}
 
 	d.peerCfg.Log.Debugf("removing lazy endpoint for peer %s", d.peerCfg.PublicKey)
+	if err := d.wgIface.RemovePeer(d.peerCfg.PublicKey); err != nil {
+		d.peerCfg.Log.Errorf("failed to remove endpoint: %s", err)
+	}
+
 	_ = d.lazyConn.Close()
 	d.bind.RemoveEndpoint(d.fakeIP)
 	d.done.Done()
 }
 
-// CapturedPacket is unused in userspace bind mode: first-packet reinjection is kernel-only.
-func (d *BindListener) CapturedPacket() []byte {
-	return nil
-}
-
 // Close stops the listener and cleans up resources.
 func (d *BindListener) Close() {
 	d.peerCfg.Log.Infof("closing activity listener (LazyConn)")

client/internal/lazyconn/activity/listener_bind_test.go

@@ -45,6 +45,10 @@ type MockWGIfaceBind struct {
 	endpointMgr *mockEndpointManager
 }
 
+func (m *MockWGIfaceBind) RemovePeer(string) error {
+	return nil
+}
+
 func (m *MockWGIfaceBind) UpdatePeer(string, []netip.Prefix, time.Duration, *net.UDPAddr, *wgtypes.Key) error {
 	return nil
 }
@@ -64,10 +68,6 @@ func (m *MockWGIfaceBind) GetBind() device.EndpointManager {
 	return m.endpointMgr
 }
 
-func (m *MockWGIfaceBind) MTU() uint16 {
-	return 1280
-}
-
 func TestBindListener_Creation(t *testing.T) {
 	mockEndpointMgr := newMockEndpointManager()
 	mockIface := &MockWGIfaceBind{endpointMgr: mockEndpointMgr}
@@ -207,9 +207,8 @@ func TestManager_BindMode(t *testing.T) {
 	require.NoError(t, err)
 
 	select {
-	case ev := <-mgr.OnActivityChan:
-		assert.Equal(t, cfg.PeerConnID, ev.PeerConnID, "Received peer connection ID should match")
-		assert.Nil(t, ev.FirstPacket, "Bind mode does not capture packets: reinjection is kernel-only")
+	case peerConnID := <-mgr.OnActivityChan:
+		assert.Equal(t, cfg.PeerConnID, peerConnID, "Received peer connection ID should match")
 	case <-time.After(2 * time.Second):
 		t.Fatal("timeout waiting for activity notification")
 	}
@@ -267,8 +266,8 @@ func TestManager_BindMode_MultiplePeers(t *testing.T) {
 	receivedPeers := make(map[peerid.ConnID]bool)
 	for i := 0; i < 2; i++ {
 		select {
-		case ev := <-mgr.OnActivityChan:
-			receivedPeers[ev.PeerConnID] = true
+		case peerConnID := <-mgr.OnActivityChan:
+			receivedPeers[peerConnID] = true
 		case <-time.After(2 * time.Second):
 			t.Fatal("timeout waiting for activity notifications")
 		}

client/internal/lazyconn/activity/listener_udp.go

@@ -3,13 +3,11 @@ package activity
 import (
 	"fmt"
 	"net"
-	"slices"
 	"sync"
 	"sync/atomic"
 
 	log "github.com/sirupsen/logrus"
 
-	"github.com/netbirdio/netbird/client/iface/bufsize"
 	"github.com/netbirdio/netbird/client/internal/lazyconn"
 )
 
@@ -22,8 +20,6 @@ type UDPListener struct {
 	done     sync.Mutex
 
 	isClosed atomic.Bool
-
-	capturedPacket []byte
 }
 
 // NewUDPListener creates a listener that detects activity via UDP socket reads.
@@ -50,13 +46,9 @@ func NewUDPListener(wgIface WgInterface, cfg lazyconn.PeerConfig) (*UDPListener,
 }
 
 // ReadPackets blocks reading from the UDP socket until activity is detected or the listener is closed.
-// The first packet that triggers activity is captured so it can be reinjected through the real
-// transport once it is established. Without this, kernel WireGuard's handshake initiation would be
-// dropped and WG would only retry after REKEY_TIMEOUT.
 func (d *UDPListener) ReadPackets() {
 	for {
-		buf := make([]byte, int(d.wgIface.MTU())+bufsize.WGBufferOverhead)
-		n, remoteAddr, err := d.conn.ReadFromUDP(buf)
+		n, remoteAddr, err := d.conn.ReadFromUDP(make([]byte, 1))
 		if err != nil {
 			if d.isClosed.Load() {
 				d.peerCfg.Log.Infof("exit from activity listener")
@@ -70,24 +62,20 @@ func (d *UDPListener) ReadPackets() {
 			d.peerCfg.Log.Warnf("received %d bytes from %s, too short", n, remoteAddr)
 			continue
 		}
-		d.capturedPacket = slices.Clone(buf[:n])
-		d.peerCfg.Log.Infof("activity detected, captured %d bytes for reinjection", n)
+		d.peerCfg.Log.Infof("activity detected")
 		break
 	}
 
-	// Leave the peer in place. ConfigureWGEndpoint will UpdatePeer with the real endpoint;
-	// removing the peer here wipes kernel WG's staged queue and drops the user packet that
-	// triggered activation.
+	d.peerCfg.Log.Debugf("removing lazy endpoint: %s", d.endpoint.String())
+	if err := d.wgIface.RemovePeer(d.peerCfg.PublicKey); err != nil {
+		d.peerCfg.Log.Errorf("failed to remove endpoint: %s", err)
+	}
+
+	// Ignore close error as it may return "use of closed network connection" if already closed.
 	_ = d.conn.Close()
 	d.done.Unlock()
 }
 
-// CapturedPacket returns the first packet that triggered activity, or nil if none was captured.
-// Safe to call after ReadPackets returns.
-func (d *UDPListener) CapturedPacket() []byte {
-	return d.capturedPacket
-}
-
 // Close stops the listener and cleans up resources.
 func (d *UDPListener) Close() {
 	d.peerCfg.Log.Infof("closing activity listener: %s", d.conn.LocalAddr().String())

client/internal/lazyconn/activity/manager.go

@@ -19,25 +19,17 @@ import (
 type listener interface {
 	ReadPackets()
 	Close()
-	CapturedPacket() []byte
-}
-
-// Event reports activity on a managed peer. FirstPacket is the bytes that triggered activation,
-// captured for reinjection through the real transport.
-type Event struct {
-	PeerConnID  peerid.ConnID
-	FirstPacket []byte
 }
 
 type WgInterface interface {
+	RemovePeer(peerKey string) error
 	UpdatePeer(peerKey string, allowedIps []netip.Prefix, keepAlive time.Duration, endpoint *net.UDPAddr, preSharedKey *wgtypes.Key) error
 	IsUserspaceBind() bool
 	Address() wgaddr.Address
-	MTU() uint16
 }
 
 type Manager struct {
-	OnActivityChan chan Event
+	OnActivityChan chan peerid.ConnID
 
 	wgIface WgInterface
 
@@ -49,7 +41,7 @@ type Manager struct {
 
 func NewManager(wgIface WgInterface) *Manager {
 	m := &Manager{
-		OnActivityChan: make(chan Event, 1),
+		OnActivityChan: make(chan peerid.ConnID, 1),
 		wgIface:        wgIface,
 		peers:          make(map[peerid.ConnID]listener),
 		done:           make(chan struct{}),
@@ -124,12 +116,12 @@ func (m *Manager) waitForTraffic(l listener, peerConnID peerid.ConnID) {
 	delete(m.peers, peerConnID)
 	m.mu.Unlock()
 
-	m.notify(Event{PeerConnID: peerConnID, FirstPacket: l.CapturedPacket()})
+	m.notify(peerConnID)
 }
 
-func (m *Manager) notify(ev Event) {
+func (m *Manager) notify(peerConnID peerid.ConnID) {
 	select {
 	case <-m.done:
-	case m.OnActivityChan <- ev:
+	case m.OnActivityChan <- peerConnID:
 	}
 }

client/internal/lazyconn/activity/manager_test.go

@@ -1,7 +1,6 @@
 package activity
 
 import (
-	"bytes"
 	"net"
 	"net/netip"
 	"testing"
@@ -26,6 +25,10 @@ func (m *MocPeer) ConnID() peerid.ConnID {
 type MocWGIface struct {
 }
 
+func (m MocWGIface) RemovePeer(string) error {
+	return nil
+}
+
 func (m MocWGIface) UpdatePeer(string, []netip.Prefix, time.Duration, *net.UDPAddr, *wgtypes.Key) error {
 	return nil
 }
@@ -41,10 +44,6 @@ func (m MocWGIface) Address() wgaddr.Address {
 	}
 }
 
-func (m MocWGIface) MTU() uint16 {
-	return 1280
-}
-
 // GetPeerListener is a test helper to access listeners
 func (m *Manager) GetPeerListener(peerConnID peerid.ConnID) (listener, bool) {
 	m.mu.Lock()
@@ -87,15 +86,11 @@ func TestManager_MonitorPeerActivity(t *testing.T) {
 	}
 
 	select {
-	case ev := <-mgr.OnActivityChan:
-		if ev.PeerConnID != peerCfg1.PeerConnID {
-			t.Fatalf("unexpected peerConnID: %v", ev.PeerConnID)
-		}
-		if !bytes.Equal(ev.FirstPacket, []byte{0x01, 0x02, 0x03, 0x04, 0x05}) {
-			t.Fatalf("unexpected first packet: %v", ev.FirstPacket)
+	case peerConnID := <-mgr.OnActivityChan:
+		if peerConnID != peerCfg1.PeerConnID {
+			t.Fatalf("unexpected peerConnID: %v", peerConnID)
 		}
 	case <-time.After(1 * time.Second):
-		t.Fatal("timed out waiting for activity")
 	}
 }
 

client/internal/lazyconn/manager/manager.go

@@ -130,8 +130,8 @@ func (m *Manager) Start(ctx context.Context) {
 		select {
 		case <-ctx.Done():
 			return
-		case ev := <-m.activityManager.OnActivityChan:
-			m.onPeerActivity(ev)
+		case peerConnID := <-m.activityManager.OnActivityChan:
+			m.onPeerActivity(peerConnID)
 		case peerIDs := <-m.inactivityManager.InactivePeersChan():
 			m.onPeerInactivityTimedOut(peerIDs)
 		}
@@ -513,13 +513,13 @@ func (m *Manager) checkHaGroupActivity(haGroup route.HAUniqueID, peerID string,
 	return false
 }
 
-func (m *Manager) onPeerActivity(ev activity.Event) {
+func (m *Manager) onPeerActivity(peerConnID peerid.ConnID) {
 	m.managedPeersMu.Lock()
 	defer m.managedPeersMu.Unlock()
 
-	mp, ok := m.managedPeersByConnID[ev.PeerConnID]
+	mp, ok := m.managedPeersByConnID[peerConnID]
 	if !ok {
-		log.Errorf("peer not found by conn id: %v", ev.PeerConnID)
+		log.Errorf("peer not found by conn id: %v", peerConnID)
 		return
 	}
 
@@ -536,7 +536,7 @@ func (m *Manager) onPeerActivity(ev activity.Event) {
 
 	m.activateHAGroupPeers(mp.peerCfg)
 
-	m.peerStore.PeerConnOpenWithFirstPacket(m.engineCtx, mp.peerCfg.PublicKey, ev.FirstPacket)
+	m.peerStore.PeerConnOpen(m.engineCtx, mp.peerCfg.PublicKey)
 }
 
 func (m *Manager) onPeerInactivityTimedOut(peerIDs map[string]struct{}) {

client/internal/lazyconn/wgiface.go

@@ -17,5 +17,4 @@ type WGIface interface {
 	IsUserspaceBind() bool
 	Address() wgaddr.Address
 	LastActivities() map[string]monotime.Time
-	MTU() uint16
 }

client/internal/mapsync.go

@@ -1,206 +0,0 @@
-package internal
-
-import (
-	"context"
-	"sync"
-	"time"
-
-	log "github.com/sirupsen/logrus"
-	"google.golang.org/protobuf/proto"
-
-	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
-)
-
-// mapStateManager is the single read/write point between the management stream
-// (writes) and the convergence loop (reads/applies).
-//
-// The stream calls SetTarget with the latest full SyncResponse — the complete
-// desired state. A single background goroutine (run) applies it to the engine in
-// bounded passes via apply() until converged, releasing syncMsgMux between passes
-// so other subsystems interleave. If a newer update arrives mid-flight, the loop
-// coalesces: it keeps converging toward the latest target and the intermediate one
-// is SKIPPED — never applied on its own (logged, no onConverged).
-//
-// Convergence is a single comparison: appliedGen == targetGen. targetGen
-// increments on every SetTarget (an internal generation counter, so it also covers
-// config-only updates that carry no network-map serial).
-//
-// onConverged fires once for each — and only each — map that is actually processed
-// (i.e. converged as the target). Skipped/superseded maps and dropped-on-error maps
-// do NOT fire it. So "sync finished in X" / RecordSyncDuration always corresponds
-// to a real, completed alignment.
-type mapStateManager struct {
-	// apply performs one bounded apply pass and reports whether more passes are needed.
-	// firstPass is true on the first pass of a given target, so the caller can run
-	// wholesale (firewall/routes/DNS/forward-rules) once per target and skip it on the
-	// re-runs that only drain the bounded peer batches. The manager owns this signal
-	// because it owns the convergence boundary; the engine need not track serials for it.
-	apply func(update *mgmProto.SyncResponse, firstPass bool) (bool, error)
-	// onConverged is called once per processed map, with the elapsed time since that
-	// map was received (for the sync-duration metric / "sync finished" log).
-	onConverged func(time.Duration)
-	// persist snapshots an update to disk for restore-on-restart. Called once per
-	// update received from management (in SetTarget), including ones later coalesced
-	// or skipped from apply, so the on-disk state mirrors what management last sent.
-	// The impl skips config-only updates (nil NetworkMap). May be nil.
-	persist func(*mgmProto.SyncResponse)
-
-	mu          sync.Mutex
-	target      *mgmProto.SyncResponse
-	targetGen   uint64
-	appliedGen  uint64
-	targetSetAt time.Time
-
-	wake chan struct{}
-}
-
-func newMapStateManager(apply func(update *mgmProto.SyncResponse, firstPass bool) (bool, error), persist func(*mgmProto.SyncResponse), onConverged func(time.Duration)) *mapStateManager {
-	return &mapStateManager{
-		apply:       apply,
-		persist:     persist,
-		onConverged: onConverged,
-		wake:        make(chan struct{}, 1),
-	}
-}
-
-// SetTarget records the latest update as the desired state and wakes the loop.
-// It returns immediately; convergence happens in the background. Serial-based
-// staleness of the network map is still enforced inside apply (updateNetworkMap).
-func (m *mapStateManager) SetTarget(update *mgmProto.SyncResponse) error {
-	m.mu.Lock()
-	// A target that has not settled yet (targetGen > appliedGen) is being superseded
-	// before it converged: we coalesce to the latest map and never apply this one on
-	// its own. It is SKIPPED — logged here, and it will not fire onConverged.
-	if m.target != nil && m.targetGen > m.appliedGen {
-		log.Debugf("sync map (gen %d) superseded before convergence, skipping", m.targetGen)
-	}
-	m.target = m.mergeTarget(m.target, update)
-	// Bump an internal generation counter, NOT the map serial: config-only updates
-	// (relay token rotation, STUN/TURN) arrive with NetworkMap == nil and carry no
-	// serial, yet must still be applied. Every SetTarget is therefore a distinct
-	// target regardless of payload. Map-serial staleness is enforced separately
-	// inside apply (updateNetworkMap).
-	m.targetGen++
-	m.targetSetAt = time.Now()
-	m.mu.Unlock()
-
-	select {
-	case m.wake <- struct{}{}:
-	default:
-	}
-
-	// Persist every update received from management — once per update (not per apply
-	// pass), and including ones that get coalesced/skipped from apply, so the on-disk
-	// state always reflects the latest map management sent. Done after waking the loop
-	// so convergence can start in parallel with the disk write. The persist impl skips
-	// config-only updates (nil NetworkMap).
-	if m.persist != nil {
-		m.persist(update)
-	}
-	return nil
-}
-
-// mergeTarget combines the currently pending target with a freshly received update
-// and returns the new desired state. It is called under m.mu from SetTarget and is
-// the single seam where the replace-vs-squash decision lives.
-//
-// Today management always sends a FULL map (the complete desired state), so the
-// update simply replaces whatever was pending — prev is ignored. When management
-// starts sending incremental/delta updates, squash `update` onto `prev` here; the
-// rest of the manager (generation tracking, convergence, signaling) is unaffected
-// because it already treats target as "the complete desired state, whatever it is".
-func (m *mapStateManager) mergeTarget(prev, update *mgmProto.SyncResponse) *mgmProto.SyncResponse {
-	// Plain replace unless a config-only update (no map) arrives while the pending
-	// target still has an unconverged map (targetGen > appliedGen). In that case graft
-	// the pending map (and its checks) onto the incoming config update, so the next pass
-	// keeps converging the map instead of settling on a nil-map target and stranding the
-	// remaining peer batches until another full map arrives.
-	if update == nil || update.GetNetworkMap() != nil || prev == nil || prev.GetNetworkMap() == nil || m.targetGen == m.appliedGen {
-		return update
-	}
-
-	merged, ok := proto.Clone(update).(*mgmProto.SyncResponse)
-	if !ok {
-		return update
-	}
-	merged.NetworkMap = prev.GetNetworkMap()
-	merged.Checks = prev.Checks
-	return merged
-}
-
-// run drives convergence until ctx is done. It is meant to run in its own goroutine.
-func (m *mapStateManager) run(ctx context.Context) {
-	// passGen is the generation of the most recent apply() call (0 = none). A pass is
-	// the first for its target when its generation differs from the previous one —
-	// true on a fresh target and on a coalesced switch to a newer target mid-flight.
-	var passGen uint64
-	for {
-		m.mu.Lock()
-		target, tg, ag := m.target, m.targetGen, m.appliedGen
-		m.mu.Unlock()
-
-		// Fully converged (or nothing yet): block until a new target arrives.
-		if target == nil || ag == tg {
-			select {
-			case <-ctx.Done():
-				return
-			case <-m.wake:
-				continue
-			}
-		}
-
-		firstPass := tg != passGen
-		passGen = tg
-		more, err := m.apply(target, firstPass)
-		if err != nil {
-			if ctx.Err() != nil {
-				return
-			}
-			// Log and DROP this target — do not retry it. A deterministic failure
-			// (e.g. a malformed peer in the map) would otherwise spin every pass
-			// making no progress. Management is the source of truth and re-delivers
-			// the full map on the next sync, so dropping is safe; peers already
-			// applied this convergence stay (idempotent diffs) and the remainder is
-			// reconciled by the next target. Mirrors the legacy handleSync path,
-			// where the apply error was logged by the gRPC client and the update
-			// dropped. No onConverged: this target did not converge.
-			log.Errorf("apply sync pass, dropping update: %v", err)
-			m.settle(tg, false)
-			continue
-		}
-
-		if more {
-			// keep converging the current target; syncMsgMux was released by apply
-			// between passes so other subsystems interleave.
-			continue
-		}
-
-		// This pass converged. Mark applied and signal this one map.
-		m.settle(tg, true)
-		// if a newer target arrived mid-pass, settle is a no-op (targetGen != tg) and
-		// ag<tg next iteration -> apply it; this generation was skipped (logged in
-		// SetTarget) and is not signaled.
-	}
-}
-
-// settle marks generation tg as processed so the loop goes idle instead of
-// re-applying the same target. It is a no-op when a newer target arrived during the
-// pass (targetGen != tg), leaving appliedGen behind so that target re-applies — the
-// just-finished generation was already counted as skipped.
-//
-// When signal is true (the pass converged) it fires onConverged once for this map;
-// when false (the target was dropped on error) it does not — the map did not converge.
-func (m *mapStateManager) settle(tg uint64, signal bool) {
-	m.mu.Lock()
-	if m.targetGen != tg {
-		m.mu.Unlock()
-		return
-	}
-	m.appliedGen = tg
-	setAt := m.targetSetAt
-	m.mu.Unlock()
-
-	if signal && m.onConverged != nil {
-		m.onConverged(time.Since(setAt))
-	}
-}

client/internal/mapsync_test.go

@@ -1,270 +0,0 @@
-package internal
-
-import (
-	"context"
-	"errors"
-	"sync/atomic"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/require"
-
-	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
-)
-
-// a config-only update arriving while a full map is still converging must keep the
-// pending map (so its remaining peer batches still apply); once converged or when the
-// pending target has no map, it replaces as usual.
-func TestMapStateManager_MergeTargetPreservesPendingMap(t *testing.T) {
-	m := newMapStateManager(nil, nil, nil)
-
-	fullMap := &mgmProto.SyncResponse{NetworkMap: &mgmProto.NetworkMap{Serial: 5}}
-	configOnly := &mgmProto.SyncResponse{NetbirdConfig: &mgmProto.NetbirdConfig{}}
-
-	// still converging the full map (targetGen > appliedGen): graft the map onto the
-	// incoming config-only update instead of dropping it
-	m.targetGen, m.appliedGen = 5, 4
-	merged := m.mergeTarget(fullMap, configOnly)
-	require.NotNil(t, merged.GetNetworkMap(), "pending map must be preserved")
-	require.EqualValues(t, 5, merged.GetNetworkMap().GetSerial())
-	require.NotNil(t, merged.GetNetbirdConfig(), "new config must be carried")
-	require.NotSame(t, configOnly, merged, "must not mutate the received update in place")
-
-	// already converged (targetGen == appliedGen): nothing pending -> plain replace
-	m.targetGen, m.appliedGen = 5, 5
-	require.Same(t, configOnly, m.mergeTarget(fullMap, configOnly))
-
-	// a full map always replaces
-	newFull := &mgmProto.SyncResponse{NetworkMap: &mgmProto.NetworkMap{Serial: 6}}
-	m.targetGen, m.appliedGen = 5, 4
-	require.Same(t, newFull, m.mergeTarget(fullMap, newFull))
-}
-
-// converges over the bounded passes (apply returns more until the 3rd pass),
-// fires onConverged exactly once, then blocks (no further apply) until a new target.
-func TestMapStateManager_ConvergesThenStops(t *testing.T) {
-	var passes int32
-	var firstPasses int32
-	converged := make(chan struct{}, 1)
-
-	apply := func(_ *mgmProto.SyncResponse, firstPass bool) (bool, error) {
-		n := atomic.AddInt32(&passes, 1)
-		if firstPass {
-			atomic.AddInt32(&firstPasses, 1)
-		}
-		return n < 3, nil // more on pass 1 and 2, converge on pass 3
-	}
-	m := newMapStateManager(apply, nil, func(time.Duration) { converged <- struct{}{} })
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-
-	select {
-	case <-converged:
-	case <-time.After(2 * time.Second):
-		t.Fatal("manager did not converge")
-	}
-	require.EqualValues(t, 3, atomic.LoadInt32(&passes))
-	require.EqualValues(t, 1, atomic.LoadInt32(&firstPasses), "firstPass true only on pass 1, false on re-runs of the same target")
-
-	// once converged the loop blocks: no further apply calls
-	time.Sleep(100 * time.Millisecond)
-	require.EqualValues(t, 3, atomic.LoadInt32(&passes), "apply must not run after convergence")
-}
-
-// persist runs once per received update (not per apply pass), regardless of how many
-// bounded passes that target takes to converge.
-func TestMapStateManager_PersistsOncePerUpdate(t *testing.T) {
-	var passes, persists int32
-	converged := make(chan struct{}, 1)
-	apply := func(_ *mgmProto.SyncResponse, _ bool) (bool, error) {
-		n := atomic.AddInt32(&passes, 1)
-		return n < 3, nil // 3 passes for one target
-	}
-	persist := func(*mgmProto.SyncResponse) { atomic.AddInt32(&persists, 1) }
-	m := newMapStateManager(apply, persist, func(time.Duration) { converged <- struct{}{} })
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	select {
-	case <-converged:
-	case <-time.After(2 * time.Second):
-		t.Fatal("did not converge")
-	}
-	require.EqualValues(t, 3, atomic.LoadInt32(&passes))
-	require.EqualValues(t, 1, atomic.LoadInt32(&persists), "persist once per update, not per pass")
-}
-
-// every update received from management is persisted — even one that is coalesced /
-// skipped from apply before it ever converges.
-func TestMapStateManager_PersistsEveryUpdateIncludingSkipped(t *testing.T) {
-	release := make(chan struct{})
-	var persists int32
-	apply := func(_ *mgmProto.SyncResponse, _ bool) (bool, error) {
-		<-release // hold the first apply so the second update coalesces/skips
-		return false, nil
-	}
-	persist := func(*mgmProto.SyncResponse) { atomic.AddInt32(&persists, 1) }
-	m := newMapStateManager(apply, persist, nil)
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{})) // map1 -> apply blocks
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{})) // map2 supersedes map1 (skipped from apply)
-	close(release)
-
-	// both updates persisted even though map1 is skipped from apply
-	require.Eventually(t, func() bool { return atomic.LoadInt32(&persists) == 2 }, 2*time.Second, 10*time.Millisecond)
-}
-
-// each map that is actually processed (converged before the next arrives) fires
-// onConverged exactly once — mirroring the legacy per-message handleSync timing.
-func TestMapStateManager_SignalsEachProcessedMap(t *testing.T) {
-	converged := make(chan struct{}, 8)
-	apply := func(_ *mgmProto.SyncResponse, _ bool) (bool, error) {
-		return false, nil // converge in one pass
-	}
-	m := newMapStateManager(apply, nil, func(time.Duration) { converged <- struct{}{} })
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	const maps = 3
-	for i := 0; i < maps; i++ {
-		require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-		select { // wait for this map to converge before sending the next (no coalescing)
-		case <-converged:
-		case <-time.After(2 * time.Second):
-			t.Fatalf("map %d not signaled", i)
-		}
-	}
-
-	// no extra signals once the stream goes quiet
-	select {
-	case <-converged:
-		t.Fatal("unexpected extra onConverged")
-	case <-time.After(100 * time.Millisecond):
-	}
-}
-
-// a map superseded before it converges is skipped: only the latest (processed) map
-// fires onConverged, not the skipped one.
-func TestMapStateManager_SkippedMapNotSignaled(t *testing.T) {
-	release := make(chan struct{})
-	var applies, converged atomic.Int32
-	apply := func(_ *mgmProto.SyncResponse, _ bool) (bool, error) {
-		applies.Add(1)
-		<-release // hold the first apply in-flight so we can queue a newer target
-		return false, nil
-	}
-	m := newMapStateManager(apply, nil, func(time.Duration) { converged.Add(1) })
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	// map1 is picked up; its apply blocks on release
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	require.Eventually(t, func() bool { return applies.Load() >= 1 }, 2*time.Second, 5*time.Millisecond)
-
-	// map2 supersedes map1 before it settled -> map1 is skipped
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	close(release) // let both applies proceed
-
-	// only the processed (latest) map signals; the skipped one does not
-	require.Eventually(t, func() bool { return converged.Load() == 1 }, 2*time.Second, 10*time.Millisecond)
-	time.Sleep(150 * time.Millisecond)
-	require.EqualValues(t, 1, converged.Load(), "skipped map must not fire onConverged")
-	require.EqualValues(t, 2, applies.Load(), "both targets entered apply (map1 once, map2 once)")
-}
-
-// an apply error drops the target: no retry of the same target, no onConverged,
-// the loop goes idle — and a fresh target is still applied afterwards.
-func TestMapStateManager_DropsTargetOnError(t *testing.T) {
-	applied := make(chan struct{}, 8)
-	var failNext atomic.Bool
-	failNext.Store(true)
-
-	apply := func(_ *mgmProto.SyncResponse, _ bool) (bool, error) {
-		applied <- struct{}{}
-		if failNext.Load() {
-			return false, errors.New("boom")
-		}
-		return false, nil // converge in one pass
-	}
-	var converged atomic.Int32
-	m := newMapStateManager(apply, nil, func(time.Duration) { converged.Add(1) })
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	// first target errors -> applied once, then dropped (no retry, no onConverged)
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	select {
-	case <-applied:
-	case <-time.After(2 * time.Second):
-		t.Fatal("errored target not applied")
-	}
-	select {
-	case <-applied:
-		t.Fatal("errored target must not be retried")
-	case <-time.After(150 * time.Millisecond):
-	}
-	require.EqualValues(t, 0, converged.Load(), "onConverged must not fire on error")
-
-	// a new target is still processed normally and converges
-	failNext.Store(false)
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	select {
-	case <-applied:
-	case <-time.After(2 * time.Second):
-		t.Fatal("new target after error not applied")
-	}
-	require.Eventually(t, func() bool { return converged.Load() == 1 }, 2*time.Second, 10*time.Millisecond)
-}
-
-// a new target after convergence triggers a fresh apply; an idle (converged)
-// manager does not apply on its own.
-func TestMapStateManager_ReappliesOnNewTarget(t *testing.T) {
-	applied := make(chan struct{}, 8)
-	apply := func(_ *mgmProto.SyncResponse, _ bool) (bool, error) {
-		applied <- struct{}{}
-		return false, nil // converge in one pass
-	}
-	m := newMapStateManager(apply, nil, nil)
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go m.run(ctx)
-
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	select {
-	case <-applied:
-	case <-time.After(2 * time.Second):
-		t.Fatal("first target not applied")
-	}
-
-	// converged → must stay idle (no spurious apply)
-	select {
-	case <-applied:
-		t.Fatal("unexpected apply while idle/converged")
-	case <-time.After(150 * time.Millisecond):
-	}
-
-	require.NoError(t, m.SetTarget(&mgmProto.SyncResponse{}))
-	select {
-	case <-applied:
-	case <-time.After(2 * time.Second):
-		t.Fatal("new target not applied")
-	}
-}

client/internal/metrics/influxdb.go

@@ -120,30 +120,6 @@ func (m *influxDBMetrics) RecordSyncDuration(_ context.Context, agentInfo AgentI
 	m.trimLocked()
 }
 
-func (m *influxDBMetrics) RecordSyncPhase(_ context.Context, agentInfo AgentInfo, phase string, duration time.Duration) {
-	tags := fmt.Sprintf("deployment_type=%s,version=%s,os=%s,arch=%s,peer_id=%s,phase=%s",
-		agentInfo.DeploymentType.String(),
-		agentInfo.Version,
-		agentInfo.OS,
-		agentInfo.Arch,
-		agentInfo.peerID,
-		phase,
-	)
-
-	m.mu.Lock()
-	defer m.mu.Unlock()
-
-	m.samples = append(m.samples, influxSample{
-		measurement: "netbird_sync_phase",
-		tags:        tags,
-		fields: map[string]float64{
-			"duration_seconds": duration.Seconds(),
-		},
-		timestamp: time.Now(),
-	})
-	m.trimLocked()
-}
-
 func (m *influxDBMetrics) RecordLoginDuration(_ context.Context, agentInfo AgentInfo, duration time.Duration, success bool) {
 	result := "success"
 	if !success {

client/internal/metrics/infra/README.md

@@ -78,25 +78,6 @@ Tags:
 - `os`: Operating system (linux, darwin, windows, android, ios, etc.)
 - `arch`: CPU architecture (amd64, arm64, etc.)
 
-### Sync Phase Timing
-
-Measurement: `netbird_sync_phase`
-
-Breaks down where time goes inside a single sync, so the total `netbird_sync` duration can be attributed to the sub-step that dominates.
-
-| Field | Description |
-|-------|-------------|
-| `duration_seconds` | Time spent in one sub-phase of sync processing |
-
-Tags:
-- `phase`: the sub-phase — `netbird_config`, `checks`, `persist`, `dns_server`, `routes_classify`, `routes_apply`, `filtering`, `dns_forwarder`, `forward_rules`, `offline_peers`, `removed_peers`, `modified_peers`, `added_peers`, `lazy_exclude`
-- `deployment_type`: "cloud" | "selfhosted" | "unknown"
-- `version`: NetBird version string
-- `os`: Operating system (linux, darwin, windows, android, ios, etc.)
-- `arch`: CPU architecture (amd64, arm64, etc.)
-
-**Note:** this is wall-time per phase — it includes both CPU work and time spent waiting on locks. A slow phase points to *where* the time goes, not *why*; pair it with lock-wait metrics to tell contention apart from real work.
-
 ### Login Duration
 
 Measurement: `netbird_login`
@@ -210,52 +191,4 @@ docker compose exec influxdb influx query \
 
 # Check ingest server health
 curl http://localhost:8087/health
-```
-
-## Analyzing a Debug Bundle
-
-Metrics collection is always on, so every debug bundle ships a `metrics.txt` in InfluxDB line protocol — a timestamped time series of all recorded events (sync durations, sync phases, connection stages, login). You can replay it into the local stack and graph it, without a running client.
-
-The bundle's `metrics.txt` is a rolling window (capped at 5 days / ~20k samples, see [Buffer Limits](#buffer-limits)). For a connection incident the relevant window is short (connection setup is seconds), so a bundle captured during the issue is enough.
-
-### 1. Start the stack
-
-```bash
-# From this directory (client/internal/metrics/infra)
-INFLUXDB_ADMIN_TOKEN=admin123 INFLUXDB_ADMIN_PASSWORD=admin123 GRAFANA_ADMIN_PASSWORD=admin123 \
-  docker compose up -d
-```
-
-(`admin123` are throwaway local credentials — fine for offline analysis.)
-
-### 2. Clear any previous data
-
-So you only see this bundle:
-
-```bash
-docker exec influxdb influx delete --org netbird --bucket metrics --token admin123 \
-  --start 1970-01-01T00:00:00Z --stop 2100-01-01T00:00:00Z
-```
-
-### 3. Import the bundle's metrics.txt
-
-InfluxDB is not exposed on the host, so import inside the container:
-
-```bash
-docker cp /path/to/bundle/metrics.txt influxdb:/tmp/m.txt
-docker exec influxdb influx write --org netbird --bucket metrics --precision ns \
-  --token admin123 --file /tmp/m.txt
-```
-
-Re-importing the same file is idempotent (same measurement+tags+timestamp overwrites).
-
-### 4. View the dashboards
-
-Grafana on http://localhost:3001 (login `admin` / `admin123`), datasource pre-provisioned:
-
-- **Where sync time goes:** http://localhost:3001/d/netbird-sync-phases/netbird-sync-phases-where-time-goes
-- **General client metrics:** http://localhost:3001/d/netbird-influxdb-metrics
-
-**Set the time range** to cover the bundle's timestamps (e.g. "Last 7 days" or an absolute range matching when the bundle was taken) — with the default short range the panels look empty.
-
-Bundles are distinguishable by the `version` tag; add a tag at import time (e.g. `sed 's/^netbird_\([a-z_]*\),/netbird_\1,bundle=mycase,/' metrics.txt`) if you want to compare several side by side.
+```

client/internal/metrics/infra/grafana/provisioning/dashboards/json/netbird-sync-phases.json

@@ -1,259 +0,0 @@
-{
-  "annotations": {
-    "list": []
-  },
-  "editable": true,
-  "fiscalYearStartMonth": 0,
-  "graphTooltip": 1,
-  "links": [],
-  "refresh": "",
-  "schemaVersion": 39,
-  "tags": [
-    "netbird",
-    "sync"
-  ],
-  "templating": {
-    "list": [
-      {
-        "current": {
-          "text": "All",
-          "value": "$__all"
-        },
-        "datasource": {
-          "type": "influxdb",
-          "uid": "influxdb"
-        },
-        "definition": "import \"influxdata/influxdb/schema\"\nschema.tagValues(bucket: \"metrics\", tag: \"version\")",
-        "includeAll": true,
-        "label": "version",
-        "multi": true,
-        "name": "version",
-        "query": "import \"influxdata/influxdb/schema\"\nschema.tagValues(bucket: \"metrics\", tag: \"version\")",
-        "refresh": 2,
-        "type": "query",
-        "allValue": ".*"
-      }
-    ]
-  },
-  "time": {
-    "from": "now-2d",
-    "to": "now"
-  },
-  "timepicker": {},
-  "timezone": "",
-  "title": "NetBird Sync Phases (where time goes)",
-  "uid": "netbird-sync-phases",
-  "version": 1,
-  "panels": [
-    {
-      "id": 1,
-      "title": "Time per phase over time (stacked, ms)",
-      "type": "timeseries",
-      "datasource": {
-        "type": "influxdb",
-        "uid": "influxdb"
-      },
-      "gridPos": {
-        "h": 10,
-        "w": 24,
-        "x": 0,
-        "y": 0
-      },
-      "fieldConfig": {
-        "defaults": {
-          "unit": "ms",
-          "custom": {
-            "drawStyle": "bars",
-            "stacking": {
-              "mode": "normal",
-              "group": "A"
-            },
-            "fillOpacity": 80,
-            "lineWidth": 0
-          }
-        },
-        "overrides": []
-      },
-      "options": {
-        "legend": {
-          "displayMode": "table",
-          "placement": "right",
-          "calcs": [
-            "max",
-            "mean"
-          ]
-        },
-        "tooltip": {
-          "mode": "multi",
-          "sort": "desc"
-        }
-      },
-      "targets": [
-        {
-          "refId": "A",
-          "datasource": {
-            "type": "influxdb",
-            "uid": "influxdb"
-          },
-          "query": "from(bucket: \"metrics\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r._measurement == \"netbird_sync_phase\" and r._field == \"duration_seconds\")\n  |> filter(fn: (r) => r.version =~ /${version:regex}/)\n  |> map(fn: (r) => ({ r with _value: r._value * 1000.0 }))\n  |> keep(columns: [\"_time\", \"_value\", \"phase\"])\n  |> group(columns: [\"phase\"])"
-        }
-      ]
-    },
-    {
-      "id": 2,
-      "title": "p95 per phase (ms)",
-      "type": "bargauge",
-      "datasource": {
-        "type": "influxdb",
-        "uid": "influxdb"
-      },
-      "gridPos": {
-        "h": 11,
-        "w": 12,
-        "x": 0,
-        "y": 10
-      },
-      "fieldConfig": {
-        "defaults": {
-          "unit": "ms",
-          "color": {
-            "mode": "continuous-GrYlRd"
-          }
-        },
-        "overrides": []
-      },
-      "options": {
-        "displayMode": "gradient",
-        "orientation": "horizontal",
-        "reduceOptions": {
-          "calcs": [
-            "lastNotNull"
-          ],
-          "fields": "",
-          "values": false
-        },
-        "showUnfilled": true
-      },
-      "targets": [
-        {
-          "refId": "A",
-          "datasource": {
-            "type": "influxdb",
-            "uid": "influxdb"
-          },
-          "query": "from(bucket: \"metrics\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r._measurement == \"netbird_sync_phase\" and r._field == \"duration_seconds\")\n  |> filter(fn: (r) => r.version =~ /${version:regex}/)\n  |> map(fn: (r) => ({ r with _value: r._value * 1000.0 }))\n  |> group(columns: [\"phase\"])\n  |> quantile(q: 0.95)\n  |> group()\n  |> sort(columns: [\"_value\"], desc: true)"
-        }
-      ]
-    },
-    {
-      "id": 3,
-      "title": "Per-phase stats (ms): mean / p95 / max",
-      "type": "table",
-      "datasource": {
-        "type": "influxdb",
-        "uid": "influxdb"
-      },
-      "gridPos": {
-        "h": 11,
-        "w": 12,
-        "x": 12,
-        "y": 10
-      },
-      "fieldConfig": {
-        "defaults": {
-          "unit": "ms"
-        },
-        "overrides": []
-      },
-      "options": {
-        "showHeader": true,
-        "sortBy": [
-          {
-            "displayName": "max",
-            "desc": true
-          }
-        ]
-      },
-      "transformations": [
-        {
-          "id": "merge",
-          "options": {}
-        }
-      ],
-      "targets": [
-        {
-          "refId": "mean",
-          "datasource": {
-            "type": "influxdb",
-            "uid": "influxdb"
-          },
-          "query": "from(bucket: \"metrics\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r._measurement == \"netbird_sync_phase\" and r._field == \"duration_seconds\")\n  |> filter(fn: (r) => r.version =~ /${version:regex}/)\n  |> map(fn: (r) => ({ r with _value: r._value * 1000.0 }))\n  |> group(columns: [\"phase\"])\n  |> mean()\n  |> group()\n  |> keep(columns: [\"phase\", \"_value\"])\n  |> rename(columns: {_value: \"mean\"})"
-        },
-        {
-          "refId": "p95",
-          "datasource": {
-            "type": "influxdb",
-            "uid": "influxdb"
-          },
-          "query": "from(bucket: \"metrics\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r._measurement == \"netbird_sync_phase\" and r._field == \"duration_seconds\")\n  |> filter(fn: (r) => r.version =~ /${version:regex}/)\n  |> map(fn: (r) => ({ r with _value: r._value * 1000.0 }))\n  |> group(columns: [\"phase\"])\n  |> quantile(q: 0.95)\n  |> group()\n  |> keep(columns: [\"phase\", \"_value\"])\n  |> rename(columns: {_value: \"p95\"})"
-        },
-        {
-          "refId": "max",
-          "datasource": {
-            "type": "influxdb",
-            "uid": "influxdb"
-          },
-          "query": "from(bucket: \"metrics\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r._measurement == \"netbird_sync_phase\" and r._field == \"duration_seconds\")\n  |> filter(fn: (r) => r.version =~ /${version:regex}/)\n  |> map(fn: (r) => ({ r with _value: r._value * 1000.0 }))\n  |> group(columns: [\"phase\"])\n  |> max()\n  |> group()\n  |> keep(columns: [\"phase\", \"_value\"])\n  |> rename(columns: {_value: \"max\"})"
-        }
-      ]
-    },
-    {
-      "id": 4,
-      "title": "Total sync duration (netbird_sync, ms) \u2014 reference",
-      "type": "timeseries",
-      "datasource": {
-        "type": "influxdb",
-        "uid": "influxdb"
-      },
-      "gridPos": {
-        "h": 8,
-        "w": 24,
-        "x": 0,
-        "y": 21
-      },
-      "fieldConfig": {
-        "defaults": {
-          "unit": "ms",
-          "custom": {
-            "drawStyle": "points",
-            "pointSize": 5
-          }
-        },
-        "overrides": []
-      },
-      "options": {
-        "legend": {
-          "displayMode": "table",
-          "placement": "right",
-          "calcs": [
-            "max",
-            "mean"
-          ]
-        },
-        "tooltip": {
-          "mode": "single"
-        }
-      },
-      "targets": [
-        {
-          "refId": "A",
-          "datasource": {
-            "type": "influxdb",
-            "uid": "influxdb"
-          },
-          "query": "from(bucket: \"metrics\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r._measurement == \"netbird_sync\" and r._field == \"duration_seconds\")\n  |> filter(fn: (r) => r.version =~ /${version:regex}/)\n  |> map(fn: (r) => ({ r with _value: r._value * 1000.0 }))\n  |> keep(columns: [\"_time\", \"_value\", \"version\"])\n  |> group(columns: [\"version\"])"
-        }
-      ]
-    }
-  ]
-}

client/internal/metrics/infra/ingest/main.go

@@ -19,7 +19,7 @@ const (
 	defaultListenAddr  = ":8087"
 	defaultInfluxDBURL = "http://influxdb:8086/api/v2/write?org=netbird&bucket=metrics&precision=ns"
 	maxBodySize        = 50 * 1024 * 1024 // 50 MB max request body
-	maxDurationSeconds = 86400.0          // reject any duration field > 24 hours
+	maxDurationSeconds = 300.0            // reject any duration field > 5 minutes
 	peerIDLength       = 16               // truncated SHA-256: 8 bytes = 16 hex chars
 	maxTagValueLength  = 64               // reject tag values longer than this
 )
@@ -59,19 +59,6 @@ var allowedMeasurements = map[string]measurementSpec{
 			"peer_id":         true,
 		},
 	},
-	"netbird_sync_phase": {
-		allowedFields: map[string]bool{
-			"duration_seconds": true,
-		},
-		allowedTags: map[string]bool{
-			"deployment_type": true,
-			"version":         true,
-			"os":              true,
-			"arch":            true,
-			"peer_id":         true,
-			"phase":           true,
-		},
-	},
 	"netbird_login": {
 		allowedFields: map[string]bool{
 			"duration_seconds": true,

client/internal/metrics/infra/ingest/main_test.go

@@ -53,14 +53,14 @@ func TestValidateLine_NegativeValue(t *testing.T) {
 }
 
 func TestValidateLine_DurationTooLarge(t *testing.T) {
-	line := `netbird_sync,deployment_type=cloud,version=1.0.0,os=linux,arch=amd64,peer_id=abc duration_seconds=100000 1234567890`
+	line := `netbird_sync,deployment_type=cloud,version=1.0.0,os=linux,arch=amd64,peer_id=abc duration_seconds=999 1234567890`
 	err := validateLine(line)
 	require.Error(t, err)
 	assert.Contains(t, err.Error(), "too large")
 }
 
 func TestValidateLine_TotalSecondsTooLarge(t *testing.T) {
-	line := `netbird_peer_connection,deployment_type=cloud,connection_type=ice,attempt_type=initial,version=1.0.0,os=linux,arch=amd64,peer_id=abc,connection_pair_id=pair total_seconds=100000 1234567890`
+	line := `netbird_peer_connection,deployment_type=cloud,connection_type=ice,attempt_type=initial,version=1.0.0,os=linux,arch=amd64,peer_id=abc,connection_pair_id=pair total_seconds=500 1234567890`
 	err := validateLine(line)
 	require.Error(t, err)
 	assert.Contains(t, err.Error(), "too large")

client/internal/metrics/metrics.go

@@ -56,9 +56,6 @@ type metricsImplementation interface {
 	// RecordSyncDuration records how long it took to process a sync message
 	RecordSyncDuration(ctx context.Context, agentInfo AgentInfo, duration time.Duration)
 
-	// RecordSyncPhase records how long a single sub-phase of sync processing took
-	RecordSyncPhase(ctx context.Context, agentInfo AgentInfo, phase string, duration time.Duration)
-
 	// RecordLoginDuration records how long the login to management took
 	RecordLoginDuration(ctx context.Context, agentInfo AgentInfo, duration time.Duration, success bool)
 
@@ -130,18 +127,6 @@ func (c *ClientMetrics) RecordSyncDuration(ctx context.Context, duration time.Du
 	c.impl.RecordSyncDuration(ctx, agentInfo, duration)
 }
 
-// RecordSyncPhase records the duration of a single sub-phase of sync processing
-func (c *ClientMetrics) RecordSyncPhase(ctx context.Context, phase string, duration time.Duration) {
-	if c == nil {
-		return
-	}
-	c.mu.RLock()
-	agentInfo := c.agentInfo
-	c.mu.RUnlock()
-
-	c.impl.RecordSyncPhase(ctx, agentInfo, phase, duration)
-}
-
 // RecordLoginDuration records how long the login to management server took
 func (c *ClientMetrics) RecordLoginDuration(ctx context.Context, duration time.Duration, success bool) {
 	if c == nil {

client/internal/metrics/push_test.go

@@ -70,9 +70,6 @@ func (m *mockMetrics) RecordConnectionStages(_ context.Context, _ AgentInfo, _ s
 func (m *mockMetrics) RecordSyncDuration(_ context.Context, _ AgentInfo, _ time.Duration) {
 }
 
-func (m *mockMetrics) RecordSyncPhase(_ context.Context, _ AgentInfo, _ string, _ time.Duration) {
-}
-
 func (m *mockMetrics) RecordLoginDuration(_ context.Context, _ AgentInfo, _ time.Duration, _ bool) {
 }
 

client/internal/peer/conn.go

@@ -6,7 +6,6 @@ import (
 	"net"
 	"net/netip"
 	"runtime"
-	"slices"
 	"sync"
 	"time"
 
@@ -137,39 +136,6 @@ type Conn struct {
 	// Connection stage timestamps for metrics
 	metricsRecorder MetricsRecorder
 	metricsStages   *MetricsStages
-
-	// pendingFirstPacket is the lazyconn-captured handshake init, replayed once the real
-	// transport is up.
-	pendingFirstPacket []byte
-}
-
-// injectPendingFirstPacket replays the captured handshake through the proxy if present, else
-// directly through the ICE conn. The packet is cleared only after a successful write, so a failed
-// or transport-less attempt leaves it available for a later reinjection. Caller must hold conn.mu.
-func (conn *Conn) injectPendingFirstPacket(proxy wgproxy.Proxy, directConn net.Conn) {
-	pkt := conn.pendingFirstPacket
-	if len(pkt) == 0 {
-		return
-	}
-
-	switch {
-	case proxy != nil:
-		if err := proxy.InjectPacket(pkt); err != nil {
-			conn.Log.Debugf("failed to reinject captured first packet via proxy: %v", err)
-			return
-		}
-	case directConn != nil:
-		if _, err := directConn.Write(pkt); err != nil {
-			conn.Log.Debugf("failed to reinject captured first packet via direct conn: %v", err)
-			return
-		}
-	default:
-		conn.Log.Debugf("no transport available to reinject captured first packet")
-		return
-	}
-
-	conn.pendingFirstPacket = nil
-	conn.Log.Debugf("reinjected captured first packet (%d bytes)", len(pkt))
 }
 
 // NewConn creates a new not opened Conn to the remote peer.
@@ -206,16 +172,6 @@ func NewConn(config ConnConfig, services ServiceDependencies) (*Conn, error) {
 // It will try to establish a connection using ICE and in parallel with relay. The higher priority connection type will
 // be used.
 func (conn *Conn) Open(engineCtx context.Context) error {
-	return conn.open(engineCtx, nil)
-}
-
-// OpenWithFirstPacket opens the connection like Open and stashes firstPacket to be replayed once
-// the real transport is established. The packet is retained only on a successful open.
-func (conn *Conn) OpenWithFirstPacket(engineCtx context.Context, firstPacket []byte) error {
-	return conn.open(engineCtx, firstPacket)
-}
-
-func (conn *Conn) open(engineCtx context.Context, firstPacket []byte) error {
 	conn.mu.Lock()
 	defer conn.mu.Unlock()
 
@@ -271,9 +227,6 @@ func (conn *Conn) open(engineCtx context.Context, firstPacket []byte) error {
 		defer conn.wg.Done()
 		conn.guard.Start(conn.ctx, conn.onGuardEvent)
 	}()
-	if len(firstPacket) > 0 {
-		conn.pendingFirstPacket = slices.Clone(firstPacket)
-	}
 	conn.opened = true
 	return nil
 }
@@ -470,8 +423,6 @@ func (conn *Conn) onICEConnectionIsReady(priority conntype.ConnPriority, iceConn
 		conn.wgProxyRelay.RedirectAs(ep)
 	}
 
-	conn.injectPendingFirstPacket(wgProxy, iceConnInfo.RemoteConn)
-
 	conn.currentConnPriority = priority
 	conn.statusICE.SetConnected()
 	conn.updateIceState(iceConnInfo, updateTime)
@@ -595,8 +546,6 @@ func (conn *Conn) onRelayConnectionIsReady(rci RelayConnInfo) {
 
 	wgConfigWorkaround()
 
-	conn.injectPendingFirstPacket(wgProxy, nil)
-
 	conn.rosenpassRemoteKey = rci.rosenpassPubKey
 	conn.currentConnPriority = conntype.Relay
 	conn.statusRelay.SetConnected()

client/internal/peer/handshaker.go

@@ -195,14 +195,14 @@ func (h *Handshaker) sendOffer() error {
 	}
 
 	offer := h.buildOfferAnswer()
-	h.log.Debugf("sending offer with serial: %s", offer.SessionIDString())
+	h.log.Infof("sending offer with serial: %s", offer.SessionIDString())
 
 	return h.signaler.SignalOffer(offer, h.config.Key)
 }
 
 func (h *Handshaker) sendAnswer() error {
 	answer := h.buildOfferAnswer()
-	h.log.Debugf("sending answer with serial: %s", answer.SessionIDString())
+	h.log.Infof("sending answer with serial: %s", answer.SessionIDString())
 
 	return h.signaler.SignalAnswer(answer, h.config.Key)
 }

client/internal/peer/status.go

@@ -7,6 +7,7 @@ import (
 	"net/netip"
 	"slices"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/google/uuid"
@@ -191,23 +192,29 @@ func (s *StatusChangeSubscription) Events() chan map[string]RouterState {
 // every private-service request) don't contend against each other.
 // Pure read methods take RLock; anything that mutates state takes Lock.
 type Status struct {
-	mux                   sync.RWMutex
-	muxRelays             sync.RWMutex
-	peers                 map[string]State
-	ipToKey               map[string]string
-	changeNotify          map[string]map[string]*StatusChangeSubscription // map[peerID]map[subscriptionID]*StatusChangeSubscription
-	signalState           bool
-	signalError           error
-	managementState       bool
-	managementError       error
-	relayStates           []relay.ProbeResult
-	localPeer             LocalPeerState
-	offlinePeers          []State
-	mgmAddress            string
-	signalAddress         string
-	notifier              *notifier
-	rosenpassEnabled      bool
-	rosenpassPermissive   bool
+	mux                 sync.RWMutex
+	peers               map[string]State
+	ipToKey             map[string]string
+	changeNotify        map[string]map[string]*StatusChangeSubscription // map[peerID]map[subscriptionID]*StatusChangeSubscription
+	signalState         bool
+	signalError         error
+	managementState     bool
+	managementError     error
+	relayStates         []relay.ProbeResult
+	localPeer           LocalPeerState
+	offlinePeers        []State
+	mgmAddress          string
+	signalAddress       string
+	notifier            *notifier
+	rosenpassEnabled    bool
+	rosenpassPermissive bool
+	// sessionExpiresAt is the absolute UTC instant at which the peer's SSO
+	// session expires. Zero when the peer is not SSO-tracked or login
+	// expiration is disabled. Populated from management LoginResponse /
+	// SyncResponse and exposed via the daemon's Status / SubscribeStatus RPC
+	// so the UI can show remaining time without itself talking to mgm.
+	sessionExpiresAt time.Time
+
 	nsGroupStates         []NSGroupState
 	resolvedDomainsStates map[domain.Domain]ResolvedDomainInfo
 	lazyConnectionEnabled bool
@@ -223,6 +230,21 @@ type Status struct {
 	eventStreams map[string]chan *proto.SystemEvent
 	eventQueue   *EventQueue
 
+	// stateChangeStreams fan-out connection-state changes (connected /
+	// disconnected / connecting / address change / peers list change) to
+	// every active SubscribeStatus gRPC stream. Each subscriber gets a
+	// buffered chan; the notifier non-blockingly pings them so a slow
+	// consumer can never stall the daemon.
+	stateChangeMux     sync.Mutex
+	stateChangeStreams map[string]chan struct{}
+
+	// networksRevision bumps whenever the routed-networks set or their
+	// selected state changes (driven by the route manager). Surfaced in the
+	// status snapshot so the UI can fingerprint on it and re-fetch
+	// ListNetworks only on a real change. Atomic so the snapshot builder can
+	// read it without taking mux.
+	networksRevision atomic.Uint64
+
 	ingressGwMgr *ingressgw.Manager
 
 	routeIDLookup routeIDLookup
@@ -237,6 +259,7 @@ func NewRecorder(mgmAddress string) *Status {
 		changeNotify:          make(map[string]map[string]*StatusChangeSubscription),
 		eventStreams:          make(map[string]chan *proto.SystemEvent),
 		eventQueue:            NewEventQueue(eventQueueSize),
+		stateChangeStreams:    make(map[string]chan struct{}),
 		offlinePeers:          make([]State, 0),
 		notifier:              newNotifier(),
 		mgmAddress:            mgmAddress,
@@ -245,8 +268,8 @@ func NewRecorder(mgmAddress string) *Status {
 }
 
 func (d *Status) SetRelayMgr(manager *relayClient.Manager) {
-	d.muxRelays.Lock()
-	defer d.muxRelays.Unlock()
+	d.mux.Lock()
+	defer d.mux.Unlock()
 	d.relayMgr = manager
 }
 
@@ -401,6 +424,7 @@ func (d *Status) UpdatePeerState(receivedState State) error {
 	if notifyRouter {
 		d.dispatchRouterPeers(receivedState.PubKey, routerSnapshot)
 	}
+	d.notifyStateChange()
 	return nil
 }
 
@@ -426,6 +450,7 @@ func (d *Status) AddPeerStateRoute(peer string, route string, resourceId route.R
 
 	// todo: consider to make sense of this notification or not
 	d.notifier.peerListChanged(numPeers)
+	d.notifyStateChange()
 	return nil
 }
 
@@ -451,6 +476,7 @@ func (d *Status) RemovePeerStateRoute(peer string, route string) error {
 
 	// todo: consider to make sense of this notification or not
 	d.notifier.peerListChanged(numPeers)
+	d.notifyStateChange()
 	return nil
 }
 
@@ -500,6 +526,7 @@ func (d *Status) UpdatePeerICEState(receivedState State) error {
 	if notifyRouter {
 		d.dispatchRouterPeers(receivedState.PubKey, routerSnapshot)
 	}
+	d.notifyStateChange()
 	return nil
 }
 
@@ -536,6 +563,7 @@ func (d *Status) UpdatePeerRelayedState(receivedState State) error {
 	if notifyRouter {
 		d.dispatchRouterPeers(receivedState.PubKey, routerSnapshot)
 	}
+	d.notifyStateChange()
 	return nil
 }
 
@@ -571,6 +599,7 @@ func (d *Status) UpdatePeerRelayedStateToDisconnected(receivedState State) error
 	if notifyRouter {
 		d.dispatchRouterPeers(receivedState.PubKey, routerSnapshot)
 	}
+	d.notifyStateChange()
 	return nil
 }
 
@@ -609,6 +638,7 @@ func (d *Status) UpdatePeerICEStateToDisconnected(receivedState State) error {
 	if notifyRouter {
 		d.dispatchRouterPeers(receivedState.PubKey, routerSnapshot)
 	}
+	d.notifyStateChange()
 	return nil
 }
 
@@ -702,6 +732,7 @@ func (d *Status) FinishPeerListModifications() {
 	for _, rd := range dispatches {
 		d.dispatchRouterPeers(rd.peerID, rd.snapshot)
 	}
+	d.notifyStateChange()
 }
 
 func (d *Status) SubscribeToPeerStateChanges(ctx context.Context, peerID string) *StatusChangeSubscription {
@@ -760,6 +791,41 @@ func (d *Status) UpdateLocalPeerState(localPeerState LocalPeerState) {
 	d.mux.Unlock()
 
 	d.notifier.localAddressChanged(fqdn, ip)
+	d.notifyStateChange()
+}
+
+// SetSessionExpiresAt records the absolute UTC instant at which the peer's
+// SSO session is set to expire. Pass the zero value to clear (e.g. when the
+// management server stops publishing a deadline because login expiration was
+// disabled or the peer is not SSO-tracked). Same-value updates are no-ops;
+// real changes fan out via notifyStateChange so SubscribeStatus consumers
+// pick up the new deadline on their next read.
+func (d *Status) SetSessionExpiresAt(deadline time.Time) {
+	d.mux.Lock()
+	if d.sessionExpiresAt.Equal(deadline) {
+		d.mux.Unlock()
+		return
+	}
+	d.sessionExpiresAt = deadline
+	d.mux.Unlock()
+	d.notifyStateChange()
+}
+
+// GetSessionExpiresAt returns the most recently recorded SSO session deadline,
+// or the zero value when no deadline is tracked. A deadline that has already
+// slipped into the past reports as "none": once the session has expired it is
+// no longer a meaningful countdown, and the sessionwatch.Watcher does not
+// arm a timer at the deadline itself to clear it (only the two pre-expiry
+// warnings). Without this guard the UI would keep painting a stale
+// "expires in …" against a moment that has passed until the next login,
+// extend, or teardown rewrote the value.
+func (d *Status) GetSessionExpiresAt() time.Time {
+	d.mux.Lock()
+	defer d.mux.Unlock()
+	if !d.sessionExpiresAt.IsZero() && d.sessionExpiresAt.Before(time.Now()) {
+		return time.Time{}
+	}
+	return d.sessionExpiresAt
 }
 
 // AddLocalPeerStateRoute adds a route to the local peer state
@@ -828,11 +894,19 @@ func (d *Status) CleanLocalPeerState() {
 	d.mux.Unlock()
 
 	d.notifier.localAddressChanged(fqdn, ip)
+	d.notifyStateChange()
 }
 
 // MarkManagementDisconnected sets ManagementState to disconnected
 func (d *Status) MarkManagementDisconnected(err error) {
 	d.mux.Lock()
+	// Health checks re-mark the same state on every probe; skip the fan-out
+	// when nothing actually changed so we don't flood SubscribeStatus
+	// consumers with identical snapshots.
+	if !d.managementState && errors.Is(d.managementError, err) {
+		d.mux.Unlock()
+		return
+	}
 	d.managementState = false
 	d.managementError = err
 	mgm := d.managementState
@@ -840,11 +914,16 @@ func (d *Status) MarkManagementDisconnected(err error) {
 	d.mux.Unlock()
 
 	d.notifier.updateServerStates(mgm, sig)
+	d.notifyStateChange()
 }
 
 // MarkManagementConnected sets ManagementState to connected
 func (d *Status) MarkManagementConnected() {
 	d.mux.Lock()
+	if d.managementState && d.managementError == nil {
+		d.mux.Unlock()
+		return
+	}
 	d.managementState = true
 	d.managementError = nil
 	mgm := d.managementState
@@ -852,6 +931,7 @@ func (d *Status) MarkManagementConnected() {
 	d.mux.Unlock()
 
 	d.notifier.updateServerStates(mgm, sig)
+	d.notifyStateChange()
 }
 
 // UpdateSignalAddress update the address of the signal server
@@ -885,6 +965,10 @@ func (d *Status) UpdateLazyConnection(enabled bool) {
 // MarkSignalDisconnected sets SignalState to disconnected
 func (d *Status) MarkSignalDisconnected(err error) {
 	d.mux.Lock()
+	if !d.signalState && errors.Is(d.signalError, err) {
+		d.mux.Unlock()
+		return
+	}
 	d.signalState = false
 	d.signalError = err
 	mgm := d.managementState
@@ -892,11 +976,16 @@ func (d *Status) MarkSignalDisconnected(err error) {
 	d.mux.Unlock()
 
 	d.notifier.updateServerStates(mgm, sig)
+	d.notifyStateChange()
 }
 
 // MarkSignalConnected sets SignalState to connected
 func (d *Status) MarkSignalConnected() {
 	d.mux.Lock()
+	if d.signalState && d.signalError == nil {
+		d.mux.Unlock()
+		return
+	}
 	d.signalState = true
 	d.signalError = nil
 	mgm := d.managementState
@@ -904,11 +993,12 @@ func (d *Status) MarkSignalConnected() {
 	d.mux.Unlock()
 
 	d.notifier.updateServerStates(mgm, sig)
+	d.notifyStateChange()
 }
 
 func (d *Status) UpdateRelayStates(relayResults []relay.ProbeResult) {
-	d.muxRelays.Lock()
-	defer d.muxRelays.Unlock()
+	d.mux.Lock()
+	defer d.mux.Unlock()
 	d.relayStates = relayResults
 }
 
@@ -1019,26 +1109,24 @@ func (d *Status) GetSignalState() SignalState {
 
 // GetRelayStates returns the stun/turn/permanent relay states
 func (d *Status) GetRelayStates() []relay.ProbeResult {
-	d.muxRelays.RLock()
+	d.mux.RLock()
+	defer d.mux.RUnlock()
 	if d.relayMgr == nil {
-		defer d.muxRelays.RUnlock()
-		return slices.Clone(d.relayStates)
+		return d.relayStates
 	}
 
-	relayMgr := d.relayMgr
 	// extend the list of stun, turn servers with the relay server connections
 	relayStates := slices.Clone(d.relayStates)
-	d.muxRelays.RUnlock()
 
-	states := relayMgr.RelayStates()
+	states := d.relayMgr.RelayStates()
 	if len(states) == 0 {
 		// no relay connection tracked yet; surface configured servers as
 		// unavailable with the real reconnect error when known
 		err := relayClient.ErrRelayClientNotConnected
-		if connErr := relayMgr.RelayConnectError(); connErr != nil {
+		if connErr := d.relayMgr.RelayConnectError(); connErr != nil {
 			err = connErr
 		}
-		for _, r := range relayMgr.ServerURLs() {
+		for _, r := range d.relayMgr.ServerURLs() {
 			relayStates = append(relayStates, relay.ProbeResult{
 				URI: r,
 				Err: err,
@@ -1110,16 +1198,19 @@ func (d *Status) GetFullStatus() FullStatus {
 // ClientStart will notify all listeners about the new service state
 func (d *Status) ClientStart() {
 	d.notifier.clientStart()
+	d.notifyStateChange()
 }
 
 // ClientStop will notify all listeners about the new service state
 func (d *Status) ClientStop() {
 	d.notifier.clientStop()
+	d.notifyStateChange()
 }
 
 // ClientTeardown will notify all listeners about the service is under teardown
 func (d *Status) ClientTeardown() {
 	d.notifier.clientTearDown()
+	d.notifyStateChange()
 }
 
 // SetConnectionListener set a listener to the notifier
@@ -1261,6 +1352,79 @@ func (d *Status) GetEventHistory() []*proto.SystemEvent {
 	return d.eventQueue.GetAll()
 }
 
+// SubscribeToStateChanges hands back a channel that receives a tick on
+// every connection-state change (connected / disconnected / connecting /
+// address change / peers-list change). The channel is buffered to one
+// pending tick so a coalesced burst still wakes the consumer exactly
+// once. Pass the returned id to UnsubscribeFromStateChanges to detach.
+func (d *Status) SubscribeToStateChanges() (string, <-chan struct{}) {
+	d.stateChangeMux.Lock()
+	defer d.stateChangeMux.Unlock()
+
+	id := uuid.New().String()
+	ch := make(chan struct{}, 1)
+	d.stateChangeStreams[id] = ch
+	return id, ch
+}
+
+// UnsubscribeFromStateChanges releases a SubscribeToStateChanges channel
+// and closes it so any consumer goroutine selecting on the channel
+// unblocks cleanly.
+func (d *Status) UnsubscribeFromStateChanges(id string) {
+	d.stateChangeMux.Lock()
+	defer d.stateChangeMux.Unlock()
+
+	if ch, ok := d.stateChangeStreams[id]; ok {
+		close(ch)
+		delete(d.stateChangeStreams, id)
+	}
+}
+
+// notifyStateChange wakes every SubscribeToStateChanges subscriber. Drops
+// the tick if a subscriber's buffer is full — by definition the consumer
+// is already going to fetch the latest snapshot, so multiple pending ticks
+// would be redundant.
+func (d *Status) notifyStateChange() {
+	d.stateChangeMux.Lock()
+	defer d.stateChangeMux.Unlock()
+
+	for _, ch := range d.stateChangeStreams {
+		select {
+		case ch <- struct{}{}:
+		default:
+		}
+	}
+}
+
+// NotifyStateChange is the public wake-the-subscribers entry point used by
+// callers that mutate state outside the peer recorder — most importantly
+// the connect-state machine, which writes StatusNeedsLogin into the
+// shared contextState (client/internal/state.go) without touching any
+// recorder field. Without this push the SubscribeStatus stream stays on
+// the previous snapshot until an unrelated peer/management/signal
+// change happens to fire notifyStateChange, leaving the UI's status
+// out of sync with the daemon.
+func (d *Status) NotifyStateChange() {
+	d.notifyStateChange()
+}
+
+// BumpNetworksRevision increments the routed-networks revision and wakes every
+// SubscribeStatus subscriber. The route manager calls it when a network map
+// changes the available routes or when a selection is applied — the peer
+// status itself only records actively-routed (chosen) networks, so without
+// this bump a candidate route appearing/disappearing would never reach the UI.
+func (d *Status) BumpNetworksRevision() {
+	d.networksRevision.Add(1)
+	d.notifyStateChange()
+}
+
+// GetNetworksRevision returns the current routed-networks revision, surfaced in
+// the status snapshot so the UI can detect route/selection changes (see
+// BumpNetworksRevision).
+func (d *Status) GetNetworksRevision() uint64 {
+	return d.networksRevision.Load()
+}
+
 func (d *Status) SetWgIface(wgInterface WGIfaceStatus) {
 	d.mux.Lock()
 	defer d.mux.Unlock()

client/internal/peer/status_test.go

@@ -314,3 +314,39 @@ func TestGetFullStatus(t *testing.T) {
 	assert.Equal(t, signalState, fullStatus.SignalState, "signal status should be equal")
 	assert.ElementsMatch(t, []State{peerState1, peerState2}, fullStatus.Peers, "peers states should match")
 }
+
+// notified reports whether a state-change tick is pending on ch, draining it.
+func notified(ch <-chan struct{}) bool {
+	select {
+	case <-ch:
+		return true
+	default:
+		return false
+	}
+}
+
+func TestMarkServerStateDoesNotNotifyWhenUnchanged(t *testing.T) {
+	status := NewRecorder("https://mgm")
+	_, ch := status.SubscribeToStateChanges()
+
+	// First transition is a real change and must notify.
+	status.MarkManagementConnected()
+	require.True(t, notified(ch), "first connect should notify")
+
+	// Re-marking the same state must not notify again.
+	status.MarkManagementConnected()
+	assert.False(t, notified(ch), "redundant connect should not notify")
+
+	// Same for signal.
+	status.MarkSignalConnected()
+	require.True(t, notified(ch), "first signal connect should notify")
+	status.MarkSignalConnected()
+	assert.False(t, notified(ch), "redundant signal connect should not notify")
+
+	// A genuine change (disconnect with an error) notifies again.
+	err := errors.New("boom")
+	status.MarkManagementDisconnected(err)
+	require.True(t, notified(ch), "disconnect should notify")
+	status.MarkManagementDisconnected(err)
+	assert.False(t, notified(ch), "redundant disconnect should not notify")
+}

client/internal/peerstore/store.go

@@ -88,24 +88,11 @@ func (s *Store) PeerConnOpen(ctx context.Context, pubKey string) {
 	if !ok {
 		return
 	}
+	// this can be blocked because of the connect open limiter semaphore
 	if err := p.Open(ctx); err != nil {
 		p.Log.Errorf("failed to open peer connection: %v", err)
 	}
-}
 
-// PeerConnOpenWithFirstPacket opens the peer connection and stashes a first packet to be
-// reinjected once the real transport is established.
-func (s *Store) PeerConnOpenWithFirstPacket(ctx context.Context, pubKey string, firstPacket []byte) {
-	s.peerConnsMu.RLock()
-	defer s.peerConnsMu.RUnlock()
-
-	p, ok := s.peerConns[pubKey]
-	if !ok {
-		return
-	}
-	if err := p.OpenWithFirstPacket(ctx, firstPacket); err != nil {
-		p.Log.Errorf("failed to open peer connection: %v", err)
-	}
 }
 
 func (s *Store) PeerConnIdle(pubKey string) {

client/internal/profilemanager/config.go

@@ -433,7 +433,7 @@ func (config *Config) apply(input ConfigInput) (updated bool, err error) {
 		updated = true
 	}
 
-	if input.ServerSSHAllowed != nil && (config.ServerSSHAllowed == nil || *input.ServerSSHAllowed != *config.ServerSSHAllowed) {
+	if input.ServerSSHAllowed != nil && *input.ServerSSHAllowed != *config.ServerSSHAllowed {
 		if *input.ServerSSHAllowed {
 			log.Infof("enabling SSH server")
 		} else {

client/internal/profilemanager/config_test.go

@@ -242,35 +242,6 @@ func TestWireguardPortDefaultVsExplicit(t *testing.T) {
 	}
 }
 
-func TestUpdateConfigServerSSHAllowedNotSet(t *testing.T) {
-	// Configs written before ServerSSHAllowed was introduced lack the field and
-	// unmarshal to nil. Supplying the SSH server flag on top of such a config must
-	// apply the value instead of panicking on a nil pointer dereference.
-	tests := []struct {
-		name  string
-		input *bool
-		want  bool
-	}{
-		{"enable", util.True(), true},
-		{"disable", util.False(), false},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			configPath := filepath.Join(t.TempDir(), "config.json")
-			require.NoError(t, os.WriteFile(configPath, []byte("{}"), 0600))
-
-			config, err := UpdateConfig(ConfigInput{
-				ConfigPath:       configPath,
-				ServerSSHAllowed: tt.input,
-			})
-			require.NoError(t, err)
-			require.NotNil(t, config.ServerSSHAllowed, "ServerSSHAllowed should be set from input")
-			assert.Equal(t, tt.want, *config.ServerSSHAllowed)
-		})
-	}
-}
-
 func TestUpdateOldManagementURL(t *testing.T) {
 	origProber := newMgmProber
 	newMgmProber = func(_ context.Context, _ string, _ wgtypes.Key, _ bool) (mgmProber, error) {

client/internal/profilemanager/state.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"os"
 	"path/filepath"
 
 	"github.com/netbirdio/netbird/util"
@@ -71,3 +72,22 @@ func (pm *ProfileManager) SetActiveProfileState(state *ProfileState) error {
 
 	return nil
 }
+
+// RemoveProfileState deletes the per-profile state file (which holds the
+// account email used for the SSO login hint and the UI display). Called after
+// a successful logout so a logged-out profile no longer shows a stale account
+// email. The state file only stores the email, so deleting it is equivalent to
+// clearing it; the next SSO login recreates it. A missing file is not an error.
+func (pm *ProfileManager) RemoveProfileState(profileName string) error {
+	configDir, err := getConfigDir()
+	if err != nil {
+		return fmt.Errorf("get config directory: %w", err)
+	}
+
+	stateFile := filepath.Join(configDir, profileName+".state.json")
+	if err := os.Remove(stateFile); err != nil && !os.IsNotExist(err) {
+		return fmt.Errorf("remove profile state: %w", err)
+	}
+
+	return nil
+}

client/internal/routemanager/dnsinterceptor/handler.go

@@ -226,11 +226,12 @@ func (d *DnsInterceptor) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 		return
 	}
 
-	// All query types for an intercepted domain are forwarded to the peer's
-	// DNS forwarder, which owns the name. Falling through to the system
-	// resolver would let it answer NXDOMAIN for a name it isn't authoritative
-	// for, poisoning the whole name (including the A/AAAA records the route
-	// does serve). The forwarder answers NODATA for types it cannot resolve.
+	// pass if non A/AAAA query
+	if r.Question[0].Qtype != dns.TypeA && r.Question[0].Qtype != dns.TypeAAAA {
+		d.continueToNextHandler(w, r, logger, "non A/AAAA query")
+		return
+	}
+
 	d.mu.RLock()
 	peerKey := d.currentPeerKey
 	d.mu.RUnlock()
@@ -250,14 +251,6 @@ func (d *DnsInterceptor) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 		r.MsgHdr.AuthenticatedData = true
 	}
 
-	// Advertise EDNS0 to the forwarder so it may return an Extended DNS Error
-	// describing why a lookup failed. The OPT is stripped from the reply when
-	// the original client did not request EDNS0.
-	hadEdns := r.IsEdns0() != nil
-	if !hadEdns {
-		r.SetEdns0(dns.DefaultMsgSize, false)
-	}
-
 	upstream := net.JoinHostPort(upstreamIP.String(), strconv.FormatUint(uint64(d.forwarderPort.Load()), 10))
 	ctx, cancel := context.WithTimeout(context.Background(), dnsTimeout)
 	defer cancel()
@@ -267,13 +260,6 @@ func (d *DnsInterceptor) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 		return
 	}
 
-	if ede, ok := resutil.ExtractEDE(reply); ok {
-		resutil.SetMeta(w, "ede", fmt.Sprintf("%d %s", ede.InfoCode, ede.ExtraText))
-	}
-	if !hadEdns {
-		resutil.StripOPT(reply)
-	}
-
 	resutil.SetMeta(w, "peer", peerKey)
 
 	reply.Id = r.Id
@@ -292,6 +278,19 @@ func (d *DnsInterceptor) writeDNSError(w dns.ResponseWriter, r *dns.Msg, logger
 	}
 }
 
+// continueToNextHandler signals the handler chain to try the next handler
+func (d *DnsInterceptor) continueToNextHandler(w dns.ResponseWriter, r *dns.Msg, logger *log.Entry, reason string) {
+	logger.Tracef("continuing to next handler for domain=%s reason=%s", r.Question[0].Name, reason)
+
+	resp := new(dns.Msg)
+	resp.SetRcode(r, dns.RcodeNameError)
+	// Set Zero bit to signal handler chain to continue
+	resp.MsgHdr.Zero = true
+	if err := w.WriteMsg(resp); err != nil {
+		logger.Errorf("failed writing DNS continue response: %v", err)
+	}
+}
+
 func (d *DnsInterceptor) getUpstreamIP(peerKey string) (netip.Addr, error) {
 	peerAllowedIP, exists := d.peerStore.AllowedIP(peerKey)
 	if !exists {