Merge branch 'feature/less-strict-metaHash' into test/affected-logic

Signed-off-by: Dmitri Dolguikh <dmitri.external@netbird.io>

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

@@ -20,7 +20,7 @@ jobs:
 
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
@@ -59,12 +59,12 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Set up Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: true

.github/workflows/git-town.yml

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

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

@@ -16,12 +16,12 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -48,7 +48,7 @@ jobs:
         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 ./... | grep -v -e /management -e /signal -e /relay -e /proxy -e /combined)
 
       - name: Upload coverage reports to Codecov
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         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@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
@@ -28,7 +28,7 @@ jobs:
         id: test
         env:
           GO_VERSION: ${{ steps.goversion.outputs.version }}
-        uses: vmactions/freebsd-vm@d1e65811565151536c0c894fff74f06351ed26e6 # v1.4.5
+        uses: vmactions/freebsd-vm@b84ab5559b5a1bb4b8ee2737d2506a16e1737636 # v1.4.8
         with:
           usesh: true
           copyback: false

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

@@ -18,7 +18,7 @@ jobs:
       management: ${{ steps.filter.outputs.management }}
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
@@ -30,7 +30,7 @@ jobs:
               - 'management/**'
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -119,12 +119,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -162,7 +162,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -175,12 +175,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -246,12 +246,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -290,7 +290,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -306,12 +306,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -347,7 +347,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -363,12 +363,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -407,7 +407,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -424,12 +424,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -484,7 +484,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         with:
           token: ${{ secrets.CODECOV_TOKEN }}
           slug: netbirdio/netbird
@@ -529,12 +529,12 @@ jobs:
             prom/prometheus
 
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -623,12 +623,12 @@ jobs:
             prom/prometheus
 
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -692,12 +692,12 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -734,7 +734,7 @@ jobs:
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'
-        uses: codecov/codecov-action@e79a6962e0d4c0c17b229090214935d2e33f8354 #v6.0.1
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f #v7.0.0
         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@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         id: go
         with:
           go-version-file: "go.mod"

.github/workflows/golangci-lint.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
       - name: codespell
@@ -40,7 +40,7 @@ jobs:
     timeout-minutes: 15
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
       - name: Check for duplicate constants
@@ -48,7 +48,7 @@ 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@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false

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

@@ -22,7 +22,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     steps:
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         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@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
       - name: Setup Android SDK
@@ -28,7 +28,7 @@ jobs:
         with:
           cmdline-tools-version: 8512546
       - name: Setup Java
-        uses: actions/setup-java@be666c2fcd27ec809703dec50e508c2fdc7f6654
+        uses: actions/setup-java@ad2b38190b15e4d6bdf0c97fb4fca8412226d287
         with:
           java-version: "11"
           distribution: "adopt"
@@ -54,11 +54,11 @@ jobs:
     runs-on: macos-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
       - name: install gomobile

.github/workflows/release.yml

@@ -27,7 +27,7 @@ jobs:
     runs-on: ubuntu-22.04
     steps:
       - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         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@d1e65811565151536c0c894fff74f06351ed26e6 # v1.4.5
+        uses: vmactions/freebsd-vm@b84ab5559b5a1bb4b8ee2737d2506a16e1737636 # v1.4.8
         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@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           fetch-depth: 0 # It is required for GoReleaser to work properly
           persist-credentials: false
@@ -166,7 +166,7 @@ jobs:
           fi
 
       - name: Set up Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -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@ce360397dd3f832beb865e1373c09c0e9f86d70a #v4.0.0
+        uses: docker/setup-qemu-action@06116385d9baf250c9f4dcb4858b16962ea869c3 #v4.1.0
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@4d04d5d9486b7bd6fa91e7baf45bbb4f8b9deedd #v4.0.0
+        uses: docker/setup-buildx-action@d7f5e7f509e45cec5c76c4d5afdd7de93d0b3df5 #v4.1.0
       - name: Login to Docker hub
         if: github.event_name != 'pull_request'
         uses: docker/login-action@650006c6eb7dba73a995cc03b0b2d7f5ca915bee # v4.2.0
@@ -221,7 +221,7 @@ jobs:
         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
       - name: Run GoReleaser
         id: goreleaser
-        uses: goreleaser/goreleaser-action@4c6ab561adb47e50c45ef534e2155934e91c40c1 # v7.2.0
+        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
         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@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           fetch-depth: 0 # It is required for GoReleaser to work properly
           persist-credentials: false
@@ -374,7 +374,7 @@ jobs:
           fi
 
       - name: Set up Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -420,7 +420,7 @@ jobs:
         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
 
       - name: Run GoReleaser
-        uses: goreleaser/goreleaser-action@4c6ab561adb47e50c45ef534e2155934e91c40c1 # v7.2.0
+        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
         with:
           version: ${{ env.GORELEASER_VER }}
           args: release --config .goreleaser_ui.yaml --clean ${{ env.flags }}
@@ -464,12 +464,12 @@ jobs:
       - if: ${{ !startsWith(github.ref, 'refs/tags/v') }}
         run: echo "flags=--snapshot" >> $GITHUB_ENV
       - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           fetch-depth: 0 # It is required for GoReleaser to work properly
           persist-credentials: false
       - name: Set up Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
           cache: false
@@ -488,7 +488,7 @@ jobs:
         run: git --no-pager diff --exit-code
       - name: Run GoReleaser
         id: goreleaser
-        uses: goreleaser/goreleaser-action@4c6ab561adb47e50c45ef534e2155934e91c40c1 # v7.2.0
+        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
         with:
           version: ${{ env.GORELEASER_VER }}
           args: release --config .goreleaser_ui_darwin.yaml --clean ${{ env.flags }}
@@ -522,7 +522,7 @@ jobs:
       downloadPath: '${{ github.workspace }}\temp'
     steps:
       - name: Checkout
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
@@ -534,13 +534,13 @@ 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@70fc10c6e5e1ce46ad2ea6f2b72d43f7d47b13c3 # v8.0.1
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
         with:
           name: release
           path: release
 
       - name: Download UI release artifacts
-        uses: actions/download-artifact@70fc10c6e5e1ce46ad2ea6f2b72d43f7d47b13c3 # v8.0.1
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1
         with:
           name: release-ui
           path: release-ui

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

@@ -68,12 +68,12 @@ jobs:
         run: sudo apt-get install -y curl
 
       - name: Checkout code
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
 
@@ -207,7 +207,7 @@ jobs:
       - name: Build management docker image
         working-directory: management
         run: |
-          docker build -t netbirdio/management:latest .
+          docker build -t netbirdio/management:latest --build-arg TARGETPLATFORM=. .
 
       - 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 .
+          docker build -t netbirdio/signal:latest --build-arg TARGETPLATFORM=. .
 
       - 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 .
+          docker build -t netbirdio/relay:latest --build-arg TARGETPLATFORM=. .
 
       - 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@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
 

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

@@ -19,11 +19,11 @@ jobs:
       GOARCH: wasm
     steps:
       - name: Checkout repository
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
       - name: Install dependencies
@@ -44,11 +44,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
         with:
           persist-credentials: false
       - name: Install Go
-        uses: actions/setup-go@4b73464bb391d4059bd26b0524d20df3927bd417 # v6.3.0
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
         with:
           go-version-file: "go.mod"
       - name: Build Wasm client

.goreleaser.yaml

@@ -462,9 +462,13 @@ 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

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: activeProf.Name,
+		ProfileName: string(activeProf.ID),
 		Username:    currUser.Username,
 	})
 	if err != nil {

client/embed/embed.go

@@ -279,9 +279,11 @@ func (c *Client) Start(startCtx context.Context) error {
 
 	select {
 	case <-startCtx.Done():
-		// 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.
+		// 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()
 		if stopErr := client.Stop(); stopErr != nil {
 			return fmt.Errorf("stop error after context done. Stop error: %w. Context done: %w", stopErr, startCtx.Err())

client/internal/connect.go

@@ -11,6 +11,7 @@ import (
 	"runtime/debug"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/cenkalti/backoff/v4"
@@ -54,6 +55,10 @@ 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
 
@@ -70,8 +75,14 @@ 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:            ctx,
+		ctx:            runCtx,
+		runCancel:      runCancel,
+		runExited:      make(chan struct{}),
 		config:         config,
 		statusRecorder: statusRecorder,
 		engineMutex:    sync.Mutex{},
@@ -135,6 +146,11 @@ 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
@@ -290,7 +306,7 @@ 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.Stop()
+				c.runCancel()
 				return backoff.Permanent(wrapErr(err)) // unrecoverable error
 			}
 			return wrapErr(err)
@@ -410,14 +426,10 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 		c.engine = nil
 		c.engineMutex.Unlock()
 
-		// 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())
+		log.Infof("ensuring wg interface is removed, Netbird engine context cancelled")
 
-			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()
 
@@ -433,12 +445,12 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 	}
 
 	c.statusRecorder.ClientStart()
-	err = backoff.Retry(operation, backOff)
+	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.Stop()
+			c.runCancel()
 		}
 		return err
 	}
@@ -516,11 +528,9 @@ func (c *ConnectClient) Status() StatusType {
 }
 
 func (c *ConnectClient) Stop() error {
-	engine := c.Engine()
-	if engine != nil {
-		if err := engine.Stop(); err != nil {
-			return fmt.Errorf("stop engine: %w", err)
-		}
+	c.runCancel()
+	if c.runStarted.Load() {
+		<-c.runExited
 	}
 	return nil
 }

client/internal/dns/mgmt/mgmt.go

@@ -51,13 +51,20 @@ type cachedRecord struct {
 }
 
 // Resolver caches critical NetBird infrastructure domains.
-// records, refreshing, mgmtDomain and serverDomains are all guarded by mutex.
+// records, refreshing, failedResolves, 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
@@ -76,9 +83,10 @@ 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),
-		cacheTTL:   resolveCacheTTL(),
+		records:        make(map[dns.Question]*cachedRecord),
+		refreshing:     make(map[dns.Question]*atomic.Bool),
+		failedResolves: make(map[domain.Domain]time.Time),
+		cacheTTL:       resolveCacheTTL(),
 	}
 }
 
@@ -173,7 +181,9 @@ 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.
+// 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.
 func (m *Resolver) AddDomain(ctx context.Context, d domain.Domain) error {
 	dnsName := strings.ToLower(dns.Fqdn(d.PunycodeString()))
 
@@ -203,6 +213,10 @@ 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
 }
 
@@ -462,6 +476,7 @@ 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
@@ -505,6 +520,7 @@ 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)
@@ -577,13 +593,85 @@ func (m *Resolver) isManagementDomain(domain domain.Domain) bool {
 	return m.mgmtDomain != nil && domain == *m.mgmtDomain
 }
 
-// addNewDomains resolves and caches all domains from the update
+// 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.
 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 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())
+		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)
 		}
 	}
 }

client/internal/dns/mgmt/mgmt_refresh_test.go

@@ -21,6 +21,7 @@ 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()
@@ -30,6 +31,7 @@ func newFakeChain() *fakeChain {
 	return &fakeChain{
 		calls:   map[string]int{},
 		answers: map[string][]dns.RR{},
+		qErr:    map[string]error{},
 		hasRoot: true,
 	}
 }
@@ -47,6 +49,9 @@ 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()
 
@@ -75,6 +80,12 @@ 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

@@ -0,0 +1,183 @@
+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

@@ -207,3 +207,35 @@ 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

@@ -120,3 +120,42 @@ func TestLookupIP_DNSErrorNotIsNotFound(t *testing.T) {
 
 	assert.Equal(t, dns.RcodeServerFailure, result.Rcode, "upstream failure should map to SERVFAIL")
 }
+
+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

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"net/netip"
 	"net/url"
+	"os"
 	"slices"
 	"strings"
 	"sync"
@@ -38,11 +39,15 @@ 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; see warningDelayFor.
-	defaultWarningDelayBase = 30 * time.Second
+	// Connected peer. Per-server and overridable via envWarningDelay;
+	// see warningDelay.
+	defaultWarningDelayBase = 60 * time.Second
 	// warningDelayBonusCap caps the route-count bonus added to the
-	// base grace window. See warningDelayFor.
+	// base grace window. See warningDelay.
 	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
@@ -135,7 +140,7 @@ type DefaultServer struct {
 	disableSys         bool
 	mux                sync.Mutex
 	service            service
-	dnsMuxMap          registeredHandlerMap
+	dnsMuxHandlers     []handlerWrapper
 	localResolver      *local.Resolver
 	wgInterface        WGIface
 	hostManager        hostManager
@@ -199,8 +204,6 @@ type handlerWrapper struct {
 	priority int
 }
 
-type registeredHandlerMap map[types.HandlerID]handlerWrapper
-
 // DefaultServerConfig holds configuration parameters for NewDefaultServer
 type DefaultServerConfig struct {
 	WgInterface    WGIface
@@ -289,7 +292,6 @@ func newDefaultServer(
 		service:           dnsService,
 		handlerChain:      handlerChain,
 		extraDomains:      make(map[domain.Domain]int),
-		dnsMuxMap:         make(registeredHandlerMap),
 		localResolver:     local.NewResolver(),
 		wgInterface:       wgInterface,
 		statusRecorder:    statusRecorder,
@@ -298,7 +300,7 @@ func newDefaultServer(
 		hostManager:       &noopHostConfigurator{},
 		mgmtCacheResolver: mgmtCacheResolver,
 		currentConfigHash: ^uint64(0), // Initialize to max uint64 to ensure first config is always applied
-		warningDelayBase:  defaultWarningDelayBase,
+		warningDelayBase:  warningDelayBaseFromEnv(),
 		healthRefresh:     make(chan struct{}, 1),
 	}
 	// Wire the local resolver against the peer status recorder so it can
@@ -328,7 +330,7 @@ func (s *DefaultServer) SetRouteSources(selected, active func() route.HAMap) {
 	type routeSettable interface {
 		setSelectedRoutes(func() route.HAMap)
 	}
-	for _, entry := range s.dnsMuxMap {
+	for _, entry := range s.dnsMuxHandlers {
 		if h, ok := entry.handler.(routeSettable); ok {
 			h.setSelectedRoutes(selected)
 		}
@@ -978,19 +980,23 @@ 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.dnsMuxMap {
+	for _, existing := range s.dnsMuxHandlers {
 		s.deregisterHandler([]string{existing.domain}, existing.priority)
-		existing.handler.Stop()
+		// 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()
+		}
 	}
 
-	muxUpdateMap := make(registeredHandlerMap)
-
 	for _, update := range muxUpdates {
 		s.registerHandler([]string{update.domain}, update.handler, update.priority)
-		muxUpdateMap[update.handler.ID()] = update
 	}
 
-	s.dnsMuxMap = muxUpdateMap
+	s.dnsMuxHandlers = muxUpdates
 }
 
 // updateNSGroupStates records the new group set and pokes the refresher.
@@ -1154,6 +1160,26 @@ 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
@@ -1204,7 +1230,7 @@ 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.dnsMuxMap {
+	for _, entry := range s.dnsMuxHandlers {
 		reporter, ok := entry.handler.(upstreamHealthReporter)
 		if !ok {
 			continue

client/internal/dns/server_test.go

@@ -104,19 +104,6 @@ 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{
@@ -132,22 +119,20 @@ func TestUpdateDNSServer(t *testing.T) {
 		},
 	}
 
-	dummyHandler := local.NewResolver()
-
 	testCases := []struct {
 		name                string
-		initUpstreamMap     registeredHandlerMap
+		initUpstreamMap     []handlerWrapper
 		initLocalZones      []nbdns.CustomZone
 		initSerial          uint64
 		inputSerial         uint64
 		inputUpdate         nbdns.Config
 		shouldFail          bool
-		expectedUpstreamMap registeredHandlerMap
+		expectedUpstreamMap []handlerWrapper
 		expectedLocalQs     []dns.Question
 	}{
 		{
 			name:            "Initial Config Should Succeed",
-			initUpstreamMap: make(registeredHandlerMap),
+			initUpstreamMap: nil,
 			initSerial:      0,
 			inputSerial:     1,
 			inputUpdate: nbdns.Config{
@@ -169,20 +154,17 @@ func TestUpdateDNSServer(t *testing.T) {
 					},
 				},
 			},
-			expectedUpstreamMap: registeredHandlerMap{
-				generateDummyHandler("netbird.io", nameServers).ID(): handlerWrapper{
+			expectedUpstreamMap: []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,
 				},
 			},
@@ -191,10 +173,10 @@ func TestUpdateDNSServer(t *testing.T) {
 		{
 			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{
+			initUpstreamMap: []handlerWrapper{
+				{
 					domain:   "netbird.cloud",
-					handler:  dummyHandler,
+					handler:  &mockHandler{},
 					priority: PriorityUpstream,
 				},
 			},
@@ -215,15 +197,13 @@ func TestUpdateDNSServer(t *testing.T) {
 					},
 				},
 			},
-			expectedUpstreamMap: registeredHandlerMap{
-				generateDummyHandler("netbird.io", nameServers).ID(): handlerWrapper{
+			expectedUpstreamMap: []handlerWrapper{
+				{
 					domain:   "netbird.io",
-					handler:  dummyHandler,
 					priority: PriorityUpstream,
 				},
-				"local-resolver": handlerWrapper{
+				{
 					domain:   "netbird.cloud",
-					handler:  dummyHandler,
 					priority: PriorityLocal,
 				},
 			},
@@ -232,7 +212,7 @@ func TestUpdateDNSServer(t *testing.T) {
 		{
 			name:            "Smaller Config Serial Should Be Skipped",
 			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: make(registeredHandlerMap),
+			initUpstreamMap: nil,
 			initSerial:      2,
 			inputSerial:     1,
 			shouldFail:      true,
@@ -240,7 +220,7 @@ func TestUpdateDNSServer(t *testing.T) {
 		{
 			name:            "Empty NS Group Domain Or Not Primary Element Should Fail",
 			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: make(registeredHandlerMap),
+			initUpstreamMap: nil,
 			initSerial:      0,
 			inputSerial:     1,
 			inputUpdate: nbdns.Config{
@@ -262,7 +242,7 @@ func TestUpdateDNSServer(t *testing.T) {
 		{
 			name:            "Invalid NS Group Nameservers list Should Fail",
 			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: make(registeredHandlerMap),
+			initUpstreamMap: nil,
 			initSerial:      0,
 			inputSerial:     1,
 			inputUpdate: nbdns.Config{
@@ -284,7 +264,7 @@ func TestUpdateDNSServer(t *testing.T) {
 		{
 			name:            "Invalid Custom Zone Records list Should Skip",
 			initLocalZones:  []nbdns.CustomZone{},
-			initUpstreamMap: make(registeredHandlerMap),
+			initUpstreamMap: nil,
 			initSerial:      0,
 			inputSerial:     1,
 			inputUpdate: nbdns.Config{
@@ -301,42 +281,41 @@ func TestUpdateDNSServer(t *testing.T) {
 					},
 				},
 			},
-			expectedUpstreamMap: registeredHandlerMap{generateDummyHandler(".", nameServers).ID(): handlerWrapper{
+			expectedUpstreamMap: []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{
+			initUpstreamMap: []handlerWrapper{
+				{
 					domain:   zoneRecords[0].Name,
-					handler:  dummyHandler,
+					handler:  &mockHandler{},
 					priority: PriorityUpstream,
 				},
 			},
 			initSerial:          0,
 			inputSerial:         1,
 			inputUpdate:         nbdns.Config{ServiceEnable: true},
-			expectedUpstreamMap: make(registeredHandlerMap),
+			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: registeredHandlerMap{
-				generateDummyHandler(zoneRecords[0].Name, nameServers).ID(): handlerWrapper{
+			initUpstreamMap: []handlerWrapper{
+				{
 					domain:   zoneRecords[0].Name,
-					handler:  dummyHandler,
+					handler:  &mockHandler{},
 					priority: PriorityUpstream,
 				},
 			},
 			initSerial:          0,
 			inputSerial:         1,
 			inputUpdate:         nbdns.Config{ServiceEnable: false},
-			expectedUpstreamMap: make(registeredHandlerMap),
+			expectedUpstreamMap: nil,
 			expectedLocalQs:     []dns.Question{},
 		},
 	}
@@ -393,7 +372,7 @@ func TestUpdateDNSServer(t *testing.T) {
 				}
 			}()
 
-			dnsServer.dnsMuxMap = testCase.initUpstreamMap
+			dnsServer.dnsMuxHandlers = testCase.initUpstreamMap
 			dnsServer.localResolver.Update(testCase.initLocalZones)
 			dnsServer.updateSerial = testCase.initSerial
 
@@ -405,14 +384,20 @@ func TestUpdateDNSServer(t *testing.T) {
 				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))
+			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 key := range testCase.expectedUpstreamMap {
-				_, found := dnsServer.dnsMuxMap[key]
+			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, key %s was not found in the dnsMuxMap: %#v", key, dnsServer.dnsMuxMap)
+					t.Fatalf("update upstream failed, handler for domain=%s priority=%d not found in dnsMuxHandlers: %#v", expected.domain, expected.priority, dnsServer.dnsMuxHandlers)
 				}
 			}
 
@@ -512,8 +497,8 @@ func TestDNSFakeResolverHandleUpdates(t *testing.T) {
 		}
 	}()
 
-	dnsServer.dnsMuxMap = registeredHandlerMap{
-		"id1": handlerWrapper{
+	dnsServer.dnsMuxHandlers = []handlerWrapper{
+		{
 			domain:   zoneRecords[0].Name,
 			handler:  &local.Resolver{},
 			priority: PriorityUpstream,
@@ -1029,15 +1014,15 @@ func (m *mockService) RegisterMux(string, dns.Handler) {}
 func (m *mockService) DeregisterMux(string)            {}
 
 func TestDefaultServer_UpdateMux(t *testing.T) {
-	baseMatchHandlers := registeredHandlerMap{
-		"upstream-group1": {
+	baseMatchHandlers := []handlerWrapper{
+		{
 			domain: "example.com",
 			handler: &mockHandler{
 				Id: "upstream-group1",
 			},
 			priority: PriorityUpstream,
 		},
-		"upstream-group2": {
+		{
 			domain: "example.com",
 			handler: &mockHandler{
 				Id: "upstream-group2",
@@ -1046,15 +1031,15 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 		},
 	}
 
-	baseRootHandlers := registeredHandlerMap{
-		"upstream-root1": {
+	baseRootHandlers := []handlerWrapper{
+		{
 			domain: ".",
 			handler: &mockHandler{
 				Id: "upstream-root1",
 			},
 			priority: PriorityDefault,
 		},
-		"upstream-root2": {
+		{
 			domain: ".",
 			handler: &mockHandler{
 				Id: "upstream-root2",
@@ -1063,22 +1048,22 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 		},
 	}
 
-	baseMixedHandlers := registeredHandlerMap{
-		"upstream-group1": {
+	baseMixedHandlers := []handlerWrapper{
+		{
 			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",
@@ -1089,7 +1074,7 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 
 	tests := []struct {
 		name             string
-		initialHandlers  registeredHandlerMap
+		initialHandlers  []handlerWrapper
 		updates          []handlerWrapper
 		expectedHandlers map[string]string // map[HandlerID]domain
 		description      string
@@ -1373,32 +1358,38 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			server := &DefaultServer{
-				dnsMuxMap:    tt.initialHandlers,
-				handlerChain: NewHandlerChain(),
-				service:      &mockService{},
+				dnsMuxHandlers: tt.initialHandlers,
+				handlerChain:   NewHandlerChain(),
+				service:        &mockService{},
 			}
 
 			// Perform the update
 			server.updateMux(tt.updates)
 
 			// Verify the results
-			assert.Equal(t, len(tt.expectedHandlers), len(server.dnsMuxMap),
+			assert.Equal(t, len(tt.expectedHandlers), len(server.dnsMuxHandlers),
 				"Number of handlers after update doesn't match expected")
 
 			// Check each expected handler
 			for id, expectedDomain := range tt.expectedHandlers {
-				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,
+				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,
 						"Domain mismatch for handler %s", id)
 				}
 			}
 
 			// Verify no unexpected handlers exist
-			for HandlerID := range server.dnsMuxMap {
-				_, expected := tt.expectedHandlers[string(HandlerID)]
-				assert.True(t, expected, "Unexpected handler found: %s", HandlerID)
+			for _, entry := range server.dnsMuxHandlers {
+				_, expected := tt.expectedHandlers[string(entry.handler.ID())]
+				assert.True(t, expected, "Unexpected handler found: %s", entry.handler.ID())
 			}
 
 			// Verify the handlerChain state and order
@@ -1413,7 +1404,7 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 
 				// Verify handler exists in mux
 				foundInMux := false
-				for _, muxEntry := range server.dnsMuxMap {
+				for _, muxEntry := range server.dnsMuxHandlers {
 					if chainEntry.Handler == muxEntry.handler &&
 						chainEntry.Priority == muxEntry.priority &&
 						chainEntry.Pattern == dns.Fqdn(muxEntry.domain) {
@@ -1422,12 +1413,108 @@ func TestDefaultServer_UpdateMux(t *testing.T) {
 					}
 				}
 				assert.True(t, foundInMux,
-					"Handler in chain not found in dnsMuxMap")
+					"Handler in chain not found in dnsMuxHandlers")
 			}
 		})
 	}
 }
 
+// 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
@@ -2049,7 +2136,6 @@ func TestBuildUpstreamHandler_MergesGroupsPerDomain(t *testing.T) {
 		localResolver: local.NewResolver(),
 		handlerChain:  NewHandlerChain(),
 		hostManager:   &noopHostConfigurator{},
-		dnsMuxMap:     make(registeredHandlerMap),
 	}
 
 	groups := []*nbdns.NameServerGroup{
@@ -2207,7 +2293,7 @@ func TestEvaluateNSGroupHealth(t *testing.T) {
 	}
 }
 
-// healthStubHandler is a minimal dnsMuxMap entry that exposes a fixed
+// healthStubHandler is a minimal dnsMuxHandlers entry that exposes a fixed
 // UpstreamHealth snapshot, letting tests drive recomputeNSGroupStates
 // without spinning up real handlers.
 type healthStubHandler struct {
@@ -2283,12 +2369,11 @@ 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.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: fx.stub, priority: PriorityUpstream}
+	fx.server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: fx.stub, priority: PriorityUpstream}}
 
 	fx.server.mux.Lock()
 	fx.server.updateNSGroupStates([]*nbdns.NameServerGroup{fx.group})
@@ -2395,7 +2480,6 @@ 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,
@@ -2407,7 +2491,7 @@ func TestProjection_OverlayAddrNoRouteDelaysWarning(t *testing.T) {
 	stub := &healthStubHandler{health: map[netip.AddrPort]UpstreamHealth{
 		overlayPeer: {LastFail: time.Now(), LastErr: "timeout"},
 	}}
-	server.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: stub, priority: PriorityUpstream}
+	server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: stub, priority: PriorityUpstream}}
 
 	server.mux.Lock()
 	server.updateNSGroupStates([]*nbdns.NameServerGroup{group})
@@ -2444,7 +2528,6 @@ 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 },
@@ -2459,7 +2542,7 @@ 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.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: stub, priority: PriorityUpstream}
+	server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: stub, priority: PriorityUpstream}}
 
 	server.mux.Lock()
 	server.updateNSGroupStates([]*nbdns.NameServerGroup{group})
@@ -2484,6 +2567,32 @@ 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
@@ -2595,7 +2704,6 @@ 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,
@@ -2613,7 +2721,7 @@ func TestProjection_MixedGroupEmitsImmediately(t *testing.T) {
 			overlay: {LastFail: time.Now(), LastErr: "timeout"},
 		},
 	}
-	server.dnsMuxMap["example.com"] = handlerWrapper{domain: "example.com", handler: stub, priority: PriorityUpstream}
+	server.dnsMuxHandlers = []handlerWrapper{{domain: "example.com", handler: stub, priority: PriorityUpstream}}
 
 	server.mux.Lock()
 	server.updateNSGroupStates([]*nbdns.NameServerGroup{group})
@@ -2640,7 +2748,6 @@ func TestDNSLoopPrevention(t *testing.T) {
 		localResolver: local.NewResolver(),
 		handlerChain:  NewHandlerChain(),
 		hostManager:   &noopHostConfigurator{},
-		dnsMuxMap:     make(registeredHandlerMap),
 	}
 
 	tests := []struct {

client/internal/dns/upstream.go

@@ -443,29 +443,32 @@ 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 {
-				stripOPT(rm)
+				resutil.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 {
-		stripOPT(rm)
+		resutil.StripOPT(rm)
 	}
 
-	u.markUpstreamOk(upstream)
 	return raceResult{msg: rm, upstream: upstream, protocol: proto}, nil
 }
 
@@ -520,22 +523,6 @@ 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

@@ -517,6 +517,78 @@ 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
@@ -913,19 +985,6 @@ 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

@@ -26,6 +26,15 @@ 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)
 }
@@ -220,7 +229,7 @@ func (f *DNSForwarder) handleDNSQuery(logger *log.Entry, w dns.ResponseWriter, q
 
 	result := resutil.LookupIP(ctx, f.resolver, network, qname, question.Qtype)
 	if result.Err != nil {
-		f.handleDNSError(ctx, logger, w, question, resp, qname, result, startTime)
+		f.handleDNSError(ctx, logger, w, question, resp, qname, result, query.IsEdns0() != nil, startTime)
 		return
 	}
 
@@ -333,6 +342,7 @@ func (f *DNSForwarder) handleDNSError(
 	resp *dns.Msg,
 	domain string,
 	result resutil.LookupResult,
+	reqHasEdns bool,
 	startTime time.Time,
 ) {
 	qType := question.Qtype
@@ -374,6 +384,10 @@ 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)
 }
 
@@ -414,3 +428,33 @@ 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

@@ -16,6 +16,7 @@ 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"
@@ -617,6 +618,85 @@ func TestDNSForwarder_ResponseCodes(t *testing.T) {
 	}
 }
 
+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")
+		})
+	}
+}
+
 func TestDNSForwarder_TCPTruncation(t *testing.T) {
 	// Test that large UDP responses are truncated with TC bit set
 	mockResolver := &MockResolver{}

client/internal/engine.go

@@ -86,6 +86,8 @@ const (
 
 var ErrResetConnection = fmt.Errorf("reset connection")
 
+var ErrEngineAlreadyStarted = errors.New("engine already started")
+
 type EngineConfig struct {
 	WgPort      int
 	WgIfaceName string
@@ -199,6 +201,8 @@ type Engine struct {
 	ctx    context.Context
 	cancel context.CancelFunc
 
+	started bool
+
 	wgInterface WGIface
 
 	udpMux *udpmux.UniversalUDPMuxDefault
@@ -279,9 +283,15 @@ 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,
@@ -314,8 +324,34 @@ 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()
 	}
@@ -366,10 +402,6 @@ func (e *Engine) Stop() error {
 	// 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()
@@ -388,21 +420,6 @@ func (e *Engine) Stop() error {
 	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.
@@ -440,18 +457,38 @@ 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) error {
+func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL) (err error) {
 	e.syncMsgMux.Lock()
 	defer e.syncMsgMux.Unlock()
 
-	if err := iface.ValidateMTU(e.config.MTU); err != nil {
+	// 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 {
 		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()
@@ -485,13 +522,11 @@ 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
@@ -526,7 +561,6 @@ 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)
 	}
 
@@ -535,7 +569,6 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 	}
 
 	if err := e.createFirewall(); err != nil {
-		e.close()
 		return err
 	}
 
@@ -547,7 +580,6 @@ 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)
 	}
 
@@ -572,9 +604,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 		e.acl = acl.NewDefaultManager(e.firewall)
 	}
 
-	err = e.dnsServer.Initialize()
-	if err != nil {
-		e.close()
+	if err := e.dnsServer.Initialize(); err != nil {
 		return fmt.Errorf("initialize dns server: %w", err)
 	}
 
@@ -586,7 +616,9 @@ 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())
 
-	e.receiveSignalEvents()
+	if err = e.receiveSignalEvents(); err != nil {
+		return err
+	}
 	e.receiveManagementEvents()
 	e.receiveJobEvents()
 
@@ -638,7 +670,6 @@ 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)
 	}
 
@@ -1698,7 +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() {
+func (e *Engine) receiveSignalEvents() error {
 	e.shutdownWg.Add(1)
 	go func() {
 		defer e.shutdownWg.Done()
@@ -1769,7 +1800,12 @@ func (e *Engine) receiveSignalEvents() {
 		}
 	}()
 
-	e.signal.WaitStreamConnected()
+	// 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
 }
 
 func (e *Engine) parseNATExternalIPMappings() []string {

client/internal/engine_test.go

@@ -247,7 +247,7 @@ func TestEngine_SSH(t *testing.T) {
 		return
 	}
 
-	ctx, cancel := context.WithCancel(context.Background())
+	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
 	defer cancel()
 
 	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
@@ -426,7 +426,7 @@ func TestEngine_UpdateNetworkMap(t *testing.T) {
 		return
 	}
 
-	ctx, cancel := context.WithCancel(context.Background())
+	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
 	defer cancel()
 
 	relayMgr := relayClient.NewManager(ctx, nil, key.PublicKey().String(), iface.DefaultMTU)
@@ -638,7 +638,7 @@ func TestEngine_Sync(t *testing.T) {
 		return
 	}
 
-	ctx, cancel := context.WithCancel(context.Background())
+	ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
 	defer cancel()
 
 	// feed updates to Engine via mocked Management client
@@ -817,7 +817,7 @@ func TestEngine_UpdateNetworkMapWithRoutes(t *testing.T) {
 				return
 			}
 
-			ctx, cancel := context.WithCancel(context.Background())
+			ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
 			defer cancel()
 
 			wgIfaceName := fmt.Sprintf("utun%d", 104+n)
@@ -1024,7 +1024,7 @@ func TestEngine_UpdateNetworkMapWithDNSUpdate(t *testing.T) {
 				return
 			}
 
-			ctx, cancel := context.WithCancel(context.Background())
+			ctx, cancel := context.WithCancel(CtxInitState(context.Background()))
 			defer cancel()
 
 			wgIfaceName := fmt.Sprintf("utun%d", 104+n)

client/internal/routemanager/dnsinterceptor/handler.go

@@ -251,6 +251,14 @@ 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()
@@ -260,6 +268,13 @@ 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

client/ios/NetBirdSDK/login.go

@@ -36,6 +36,7 @@ type URLOpener interface {
 // Auth can register or login new client
 type Auth struct {
 	ctx     context.Context
+	cancel  context.CancelFunc
 	config  *profilemanager.Config
 	cfgPath string
 }
@@ -51,8 +52,19 @@ func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
 		return nil, err
 	}
 
+	// Use a cancellable context so Stop() can abort an in-progress interactive
+	// login. The PKCE flow's WaitToken blocks (and keeps its loopback HTTP server
+	// bound to a port) until the OAuth callback arrives or the flow expires;
+	// cancelling the context unblocks WaitToken, which then shuts that server down
+	// and frees the port for the next login attempt. iOS runs login in the main-app
+	// process (decoupled from the network extension), so without this the server
+	// lingers after the user dismisses the browser and the next connect stalls
+	// trying to bind the same port.
+	ctx, cancel := context.WithCancel(context.Background())
+
 	return &Auth{
-		ctx:     context.Background(),
+		ctx:     ctx,
+		cancel:  cancel,
 		config:  cfg,
 		cfgPath: cfgPath,
 	}, nil
@@ -60,12 +72,24 @@ func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
 
 // NewAuthWithConfig instantiate Auth based on existing config
 func NewAuthWithConfig(ctx context.Context, config *profilemanager.Config) *Auth {
+	ctx, cancel := context.WithCancel(ctx)
 	return &Auth{
 		ctx:    ctx,
+		cancel: cancel,
 		config: config,
 	}
 }
 
+// Stop aborts an in-progress interactive login started via Login/LoginWithDeviceName.
+// It cancels the auth context, which unblocks the PKCE WaitToken and shuts down its
+// loopback HTTP server, freeing the redirect port. Safe to call multiple times and
+// safe to call when no login is running.
+func (a *Auth) Stop() {
+	if a.cancel != nil {
+		a.cancel()
+	}
+}
+
 // SaveConfigIfSSOSupported test the connectivity with the management server by retrieving the server device flow info.
 // If it returns a flow info than save the configuration and return true. If it gets a codes.NotFound, it means that SSO
 // is not supported and returns false without saving the configuration. For other errors return false.

client/server/server.go

@@ -993,6 +993,10 @@ func (s *Server) cleanupConnection() error {
 		return nil
 	}
 
+	// TODO: consider calling s.connectClient.Stop() instead of engine.Stop().
+	// actCancel() lets the run loop stop the engine too, so both stop it
+	// concurrently; ConnectClient.Stop cancels and waits for the run loop,
+	// making the run loop the sole owner of engine shutdown.
 	if engine != nil {
 		if err := engine.Stop(); err != nil {
 			return err

infrastructure_files/getting-started-enterprise.sh

@@ -0,0 +1,616 @@
+#!/bin/bash
+
+set -e
+set -o pipefail
+
+# NetBird Enterprise — Getting Started
+# Single-node bootstrap for a self-hosted NetBird Enterprise stack with the
+# embedded identity provider. Owner is created via first-login flow.
+
+SED_STRIP_PADDING='s/=//g'
+
+check_docker_compose() {
+  if command -v docker-compose &> /dev/null; then
+    echo "docker-compose"
+    return
+  fi
+  if docker compose --help &> /dev/null; then
+    echo "docker compose"
+    return
+  fi
+  echo "docker-compose is not installed or not in PATH. See https://docs.docker.com/engine/install/" > /dev/stderr
+  exit 1
+}
+
+check_openssl() {
+  if ! command -v openssl &> /dev/null; then
+    echo "openssl is not installed or not in PATH." > /dev/stderr
+    exit 1
+  fi
+}
+
+rand_secret() {
+  openssl rand -base64 32 | sed "$SED_STRIP_PADDING"
+}
+
+rand_b64_key() {
+  openssl rand -base64 32
+}
+
+check_nb_domain() {
+  local domain="$1"
+  if [[ -z "$domain" ]]; then
+    echo "The domain cannot be empty." > /dev/stderr
+    return 1
+  fi
+  if [[ "$domain" == "netbird.example.com" ]]; then
+    echo "The domain cannot be netbird.example.com" > /dev/stderr
+    return 1
+  fi
+  if [[ "$domain" =~ ^[0-9.]+$ ]]; then
+    echo "An IP address is not allowed. A real DNS-resolvable domain is required for TLS and the embedded IdP issuer." > /dev/stderr
+    return 1
+  fi
+  if [[ ! "$domain" =~ ^[A-Za-z0-9]([A-Za-z0-9-]*[A-Za-z0-9])?(\.[A-Za-z0-9]([A-Za-z0-9-]*[A-Za-z0-9])?)+$ ]]; then
+    echo "The value '$domain' is not a valid FQDN. A real DNS-resolvable domain is required for TLS and the embedded IdP issuer." > /dev/stderr
+    return 1
+  fi
+  return 0
+}
+
+check_domain_resolves() {
+  local domain="$1"
+  if command -v getent &> /dev/null && getent hosts "$domain" &> /dev/null; then return 0; fi
+  if command -v host &> /dev/null && host "$domain" &> /dev/null; then return 0; fi
+  if command -v dig &> /dev/null && [[ -n "$(dig +short "$domain" 2>/dev/null)" ]]; then return 0; fi
+  if command -v nslookup &> /dev/null && nslookup "$domain" &> /dev/null; then return 0; fi
+  return 1
+}
+
+read_nb_domain() {
+  local value=""
+  echo -n "Enter the FQDN for NetBird (must resolve via DNS, e.g. netbird.my-domain.com): " > /dev/stderr
+  read -r value < /dev/tty
+  if ! check_nb_domain "$value"; then
+    read_nb_domain
+    return
+  fi
+  if ! check_domain_resolves "$value"; then
+    echo "" > /dev/stderr
+    echo "Warning: '$value' does not resolve via DNS from this host." > /dev/stderr
+    echo "Caddy will not be able to issue TLS certificates until it does." > /dev/stderr
+    local confirm=""
+    echo -n "Continue anyway? [y/N]: " > /dev/stderr
+    read -r confirm < /dev/tty
+    if [[ ! "$confirm" =~ ^[Yy]$ ]]; then
+      read_nb_domain
+      return
+    fi
+  fi
+  echo "$value"
+}
+
+read_required() {
+  local prompt="$1"
+  local value=""
+  while [[ -z "$value" ]]; do
+    echo -n "$prompt: " > /dev/stderr
+    read -r value < /dev/tty
+    if [[ -z "$value" ]]; then
+      echo "Value cannot be empty." > /dev/stderr
+    fi
+  done
+  echo "$value"
+}
+
+read_secret() {
+  local prompt="$1"
+  local value=""
+  while [[ -z "$value" ]]; do
+    echo -n "$prompt: " > /dev/stderr
+    read -rs value < /dev/tty
+    echo "" > /dev/stderr
+    if [[ -z "$value" ]]; then
+      echo "Value cannot be empty." > /dev/stderr
+    fi
+  done
+  echo "$value"
+}
+
+# read_yes_no "<prompt>" [<default y|n>]
+read_yes_no() {
+  local prompt="$1"
+  local default="${2:-n}"
+  local hint
+  if [[ "$default" == "y" ]]; then
+    hint="[Y/n]"
+  else
+    hint="[y/N]"
+  fi
+  echo -n "${prompt} ${hint}: " > /dev/stderr
+  local ans=""
+  read -r ans < /dev/tty
+  if [[ -z "$ans" ]]; then
+    ans="$default"
+  fi
+  case "$ans" in
+    [Yy] | [Yy][Ee][Ss]) echo "yes" ;;
+    *) echo "no" ;;
+  esac
+}
+
+wait_postgres() {
+  set +e
+  echo -n "Waiting for postgres to become ready"
+  local counter=1
+  while true; do
+    if $DOCKER_COMPOSE_COMMAND exec -T postgres pg_isready -U "$POSTGRES_USER" -d "$POSTGRES_DB" &> /dev/null; then
+      break
+    fi
+    if [[ $counter -eq 60 ]]; then
+      echo ""
+      echo "Postgres is taking too long. Recent logs:"
+      $DOCKER_COMPOSE_COMMAND logs --tail=20 postgres
+      exit 1
+    fi
+    echo -n " ."
+    sleep 2
+    counter=$((counter + 1))
+  done
+  echo " done"
+  set -e
+}
+
+init_environment() {
+  check_openssl
+  DOCKER_COMPOSE_COMMAND=$(check_docker_compose)
+
+  if [[ -f .env ]] || [[ -f docker-compose.yml ]] || [[ -f config.yaml ]] || [[ -f Caddyfile ]]; then
+    echo "Generated files already exist in $(pwd)."
+    echo "If you want to reinitialize the environment, please remove them first:"
+    echo "  $DOCKER_COMPOSE_COMMAND down --volumes # removes all containers and volumes"
+    echo "  rm -f .env docker-compose.yml Caddyfile config.yaml"
+    echo "Be aware this will remove all data from the database."
+    exit 1
+  fi
+
+  echo "NetBird Enterprise bootstrap"
+  echo ""
+  echo "Traffic flow:"
+  echo "  Enables traffic events logging on the management server."
+  echo "  When enabled, the NetBird stack also runs NATS along with two"
+  echo "  additional containers: netbird-receiver (the traffic log receiver"
+  echo "  service) and netbird-enricher (the traffic log enricher service)."
+  echo "  It still has to be turned on from the dashboard settings afterwards."
+  echo "  See https://docs.netbird.io/manage/activity/traffic-events-logging"
+  NETBIRD_TRAFFIC_FLOW=$(read_yes_no "Enable traffic flow" "n")
+
+  echo ""
+  NETBIRD_DOMAIN=$(read_nb_domain)
+
+  echo ""
+
+  NETBIRD_LICENSE_KEY=$(read_secret "Enter license key (input hidden)")
+
+  GHCR_USERNAME="netbirdExtAccess1"
+  GHCR_TOKEN=$(read_secret "Enter GHCR token (input hidden)")
+
+  POSTGRES_USER="netbird"
+  POSTGRES_DB="netbird"
+  POSTGRES_PASSWORD=$(rand_secret)
+  NETBIRD_ENCRYPTION_KEY=$(rand_b64_key)
+  NETBIRD_RELAY_AUTH_SECRET=$(rand_secret)
+
+  POSTGRES_DSN="host=postgres user=${POSTGRES_USER} password=${POSTGRES_PASSWORD} dbname=${POSTGRES_DB} port=5432 sslmode=disable TimeZone=UTC"
+  NETBIRD_RELAY_ENDPOINT="rels://${NETBIRD_DOMAIN}:443"
+
+  echo ""
+  echo "Selected:"
+  echo "  Traffic flow: ${NETBIRD_TRAFFIC_FLOW}"
+  echo "  Domain:       ${NETBIRD_DOMAIN}"
+  echo ""
+  echo "Rendering files into $(pwd) ..."
+  install -m 600 /dev/null .env
+  render_env >> .env
+  render_docker_compose > docker-compose.yml
+
+  if [[ -z "${NETBIRD_LICENSE_SERVER_BASE_URL:-}" ]]; then
+    sed -i.bak '/NETBIRD_LICENSE_SERVER_BASE_URL/d' docker-compose.yml && rm -f docker-compose.yml.bak
+  fi
+  render_caddyfile > Caddyfile
+  install -m 600 /dev/null config.yaml
+  render_config_yaml >> config.yaml
+
+  echo "Logging in to ghcr.io ..."
+  printf '%s' "$GHCR_TOKEN" | docker login ghcr.io -u "$GHCR_USERNAME" --password-stdin
+  unset GHCR_TOKEN
+
+  echo ""
+  echo "Pulling images ..."
+  $DOCKER_COMPOSE_COMMAND pull
+
+  echo ""
+  echo "Starting postgres ..."
+  $DOCKER_COMPOSE_COMMAND up -d postgres
+  sleep 2
+  wait_postgres
+
+  echo ""
+  echo "Starting remaining services ..."
+  $DOCKER_COMPOSE_COMMAND up -d
+
+  echo ""
+  echo "Done."
+  echo ""
+  echo "Dashboard: https://${NETBIRD_DOMAIN}"
+  echo ""
+  echo "Open the dashboard in a browser to complete the first-login owner setup."
+  echo "All configuration and secrets are stored (mode 600) in $(pwd)/.env"
+  echo ""
+  echo "Tail logs:"
+  echo "  cd $(pwd) && $DOCKER_COMPOSE_COMMAND logs -f netbird-server caddy"
+}
+
+# ------------------------------------------------------------------
+# Renderers
+# ------------------------------------------------------------------
+
+render_env() {
+  cat <<EOF
+# Generated by getting-started-enterprise.sh
+# Holds all configuration and secrets for the stack. Mode 600.
+
+# Features (set by the script; don't edit without re-running)
+NETBIRD_TRAFFIC_FLOW_ENABLED=${NETBIRD_TRAFFIC_FLOW}
+
+# Domain
+NETBIRD_DOMAIN=${NETBIRD_DOMAIN}
+
+# Image tags. Default to "latest"
+NETBIRD_DASHBOARD_TAG=${NETBIRD_DASHBOARD_TAG:-latest}
+NETBIRD_SERVER_TAG=${NETBIRD_SERVER_TAG:-latest}
+EOF
+
+  if [[ "$NETBIRD_TRAFFIC_FLOW" == "yes" ]]; then
+    cat <<EOF
+NETBIRD_ENRICHER_TAG=${NETBIRD_ENRICHER_TAG:-latest}
+NETBIRD_RECEIVER_TAG=${NETBIRD_RECEIVER_TAG:-latest}
+EOF
+  fi
+
+  cat <<EOF
+
+# License keys
+EOF
+  if [[ -n "${NETBIRD_LICENSE_SERVER_BASE_URL:-}" ]]; then
+    cat <<EOF
+NETBIRD_LICENSE_SERVER_BASE_URL=${NETBIRD_LICENSE_SERVER_BASE_URL}
+EOF
+  fi
+  cat <<EOF
+NETBIRD_LICENSE_KEY=${NETBIRD_LICENSE_KEY}
+EOF
+
+  cat <<EOF
+
+# Postgres
+POSTGRES_USER=${POSTGRES_USER}
+POSTGRES_DB=${POSTGRES_DB}
+POSTGRES_PASSWORD=${POSTGRES_PASSWORD}
+NETBIRD_STORE_ENGINE_POSTGRES_DSN=${POSTGRES_DSN}
+
+# Relay
+NETBIRD_RELAY_ENDPOINT=${NETBIRD_RELAY_ENDPOINT}
+NETBIRD_RELAY_AUTH_SECRET=${NETBIRD_RELAY_AUTH_SECRET}
+
+# Datastore encryption
+NETBIRD_ENCRYPTION_KEY=${NETBIRD_ENCRYPTION_KEY}
+
+# Dashboard OIDC scopes
+NETBIRD_AUTH_SUPPORTED_SCOPES=${NETBIRD_AUTH_SUPPORTED_SCOPES:-openid profile email groups}
+EOF
+}
+
+render_docker_compose() {
+  render_compose_header
+  render_compose_common
+  render_compose_server
+  if [[ "$NETBIRD_TRAFFIC_FLOW" == "yes" ]]; then
+    render_compose_flow
+  fi
+  render_compose_postgres
+  render_compose_footer
+}
+
+render_compose_header() {
+  cat <<'EOF'
+x-default: &default
+  restart: unless-stopped
+  logging:
+    driver: json-file
+    options:
+      max-size: '500m'
+      max-file: '2'
+
+services:
+EOF
+}
+
+render_compose_common() {
+  cat <<'EOF'
+  caddy:
+    <<: *default
+    image: caddy:2
+    container_name: netbird-caddy
+    networks: [netbird]
+    environment:
+      - CADDY_SECURE_DOMAIN=${NETBIRD_DOMAIN}
+    ports:
+      - '443:443'
+      - '443:443/udp'
+      - '80:80'
+    volumes:
+      - netbird_caddy_data:/data
+      - ./Caddyfile:/etc/caddy/Caddyfile
+
+  dashboard:
+    <<: *default
+    image: ghcr.io/netbirdio/dashboard-cloud:${NETBIRD_DASHBOARD_TAG}
+    container_name: netbird-dashboard
+    networks: [netbird]
+    environment:
+      - NETBIRD_MGMT_API_ENDPOINT=https://${NETBIRD_DOMAIN}
+      - NETBIRD_MGMT_GRPC_API_ENDPOINT=https://${NETBIRD_DOMAIN}
+      - AUTH_AUDIENCE=netbird-dashboard
+      - AUTH_CLIENT_ID=netbird-dashboard
+      - AUTH_CLIENT_SECRET=
+      - AUTH_AUTHORITY=https://${NETBIRD_DOMAIN}/oauth2
+      - USE_AUTH0=false
+      - AUTH_SUPPORTED_SCOPES=${NETBIRD_AUTH_SUPPORTED_SCOPES}
+      - AUTH_REDIRECT_URI=/nb-auth
+      - AUTH_SILENT_REDIRECT_URI=/nb-silent-auth
+      - NETBIRD_TOKEN_SOURCE=accessToken
+      - NGINX_SSL_PORT=443
+      - LETSENCRYPT_DOMAIN=
+      - LETSENCRYPT_EMAIL=
+
+EOF
+}
+
+render_compose_server() {
+  cat <<'EOF'
+  netbird-server:
+    <<: *default
+    image: ghcr.io/netbirdio/netbird-server-cloud:${NETBIRD_SERVER_TAG}
+    container_name: netbird-server
+    networks: [netbird]
+    depends_on:
+      dashboard:
+        condition: service_started
+      postgres:
+        condition: service_healthy
+    ports:
+      - '3478:3478/udp'
+    volumes:
+      - netbird_data:/var/lib/netbird
+      - ./config.yaml:/etc/netbird/config.yaml
+    command: ["--config", "/etc/netbird/config.yaml"]
+    environment:
+      - NB_LICENSE_KEY=${NETBIRD_LICENSE_KEY}
+      - NETBIRD_LICENSE_SERVER_BASE_URL=${NETBIRD_LICENSE_SERVER_BASE_URL}
+
+EOF
+}
+
+render_compose_flow() {
+  cat <<'EOF'
+  nats:
+    <<: *default
+    image: nats:2
+    container_name: netbird-nats
+    networks: [netbird]
+    volumes:
+      - netbird_nats_data:/data
+    command: ["-m", "8222", "--jetstream", "--store_dir", "/data"]
+
+  enricher:
+    <<: *default
+    image: ghcr.io/netbirdio/flow-enricher-cloud:${NETBIRD_ENRICHER_TAG}
+    container_name: netbird-enricher
+    networks: [netbird]
+    depends_on:
+      postgres:
+        condition: service_healthy
+      nats:
+        condition: service_started
+    volumes:
+      - netbird_enricher:/var/lib/netbird
+    environment:
+      - NB_LICENSE_KEY=${NETBIRD_LICENSE_KEY}
+      - NETBIRD_LICENSE_SERVER_BASE_URL=${NETBIRD_LICENSE_SERVER_BASE_URL}
+      - NB_DATADIR=/var/lib/netbird
+      - NB_MANAGEMENT_STORE_ENGINE=postgres
+      - NB_MANAGEMENT_POSTGRES_DSN=${NETBIRD_STORE_ENGINE_POSTGRES_DSN}
+      - NETBIRD_STORE_ENGINE_POSTGRES_DSN=${NETBIRD_STORE_ENGINE_POSTGRES_DSN}
+      - NB_TRAFFIC_EVENT_POSTGRES_DSN=${NETBIRD_STORE_ENGINE_POSTGRES_DSN}
+      - NB_TRAFFIC_EVENT_STORE_ENGINE=postgres
+      - NB_MANAGEMENT_STORE_KEY=${NETBIRD_ENCRYPTION_KEY}
+      - NB_FLOW_ADAPTER_TYPE=nats
+      - NB_FLOW_NATS_ENDPOINTS=nats://nats:4222
+      - NB_FLOW_NATS_STREAM=traffic-events
+      - NB_METRICS_PORT=9091
+      - NB_PERSISTENCE_RETENTION_PERIOD=168h
+
+  receiver:
+    <<: *default
+    image: ghcr.io/netbirdio/flow-receiver-cloud:${NETBIRD_RECEIVER_TAG}
+    container_name: netbird-receiver
+    networks: [netbird]
+    depends_on:
+      nats:
+        condition: service_started
+    environment:
+      - NB_LICENSE_KEY=${NETBIRD_LICENSE_KEY}
+      - NETBIRD_LICENSE_SERVER_BASE_URL=${NETBIRD_LICENSE_SERVER_BASE_URL}
+      - NB_FLOW_LISTEN_PORT=80
+      - NB_FLOW_ADAPTER_TYPE=nats
+      - NB_FLOW_NATS_ENDPOINTS=nats://nats:4222
+      - NB_FLOW_NATS_STREAM=traffic-events
+      - NB_FLOW_AUTH_SECRET=${NETBIRD_RELAY_AUTH_SECRET}
+
+EOF
+}
+
+render_compose_postgres() {
+  cat <<'EOF'
+  postgres:
+    <<: *default
+    image: postgres:17
+    container_name: netbird-postgres
+    networks: [netbird]
+    environment:
+      - POSTGRES_USER=${POSTGRES_USER}
+      - POSTGRES_PASSWORD=${POSTGRES_PASSWORD}
+      - POSTGRES_DB=${POSTGRES_DB}
+    healthcheck:
+      test: ["CMD-SHELL", "pg_isready -U ${POSTGRES_USER} -d ${POSTGRES_DB}"]
+      interval: 10s
+      timeout: 5s
+      retries: 10
+    volumes:
+      - netbird_postgres:/var/lib/postgresql/data
+
+EOF
+}
+
+render_compose_footer() {
+  cat <<'EOF'
+volumes:
+  netbird_data:
+EOF
+  if [[ "$NETBIRD_TRAFFIC_FLOW" == "yes" ]]; then
+    cat <<'EOF'
+  netbird_nats_data:
+  netbird_enricher:
+EOF
+  fi
+  cat <<'EOF'
+  netbird_postgres:
+  netbird_caddy_data:
+
+networks:
+  netbird:
+EOF
+}
+
+render_caddyfile() {
+  cat <<'EOF'
+{
+  servers :80,:443 {
+    protocols h1 h2c h2 h3
+  }
+}
+
+(security_headers) {
+    header * {
+        Strict-Transport-Security "max-age=3600; includeSubDomains; preload"
+        X-Content-Type-Options "nosniff"
+        X-Frame-Options "SAMEORIGIN"
+        X-XSS-Protection "1; mode=block"
+        -Server
+        Referrer-Policy strict-origin-when-cross-origin
+    }
+}
+
+:80 {
+    redir https://{$CADDY_SECURE_DOMAIN}{uri} permanent
+}
+
+{$CADDY_SECURE_DOMAIN}:443 {
+    import security_headers
+    # Signal (gRPC over h2c)
+    reverse_proxy /signalexchange.SignalExchange/* h2c://netbird-server:80
+    # Management (gRPC over h2c + HTTP)
+    reverse_proxy /management.ManagementService/* h2c://netbird-server:80
+    reverse_proxy /api/* netbird-server:80
+    reverse_proxy /ws-proxy/* netbird-server:80
+    # Embedded IdP (OAuth2 endpoints served by netbird server)
+    reverse_proxy /oauth2/* netbird-server:80
+    # Relay (WebSocket multiplexed on the same port)
+    reverse_proxy /relay* netbird-server:80
+EOF
+
+  if [[ "$NETBIRD_TRAFFIC_FLOW" == "yes" ]]; then
+    cat <<'EOF'
+    # Flow receiver (gRPC over h2c)
+    reverse_proxy /flow.FlowService/* h2c://receiver:80
+EOF
+  fi
+
+  cat <<'EOF'
+    # Dashboard
+    reverse_proxy /* dashboard:80
+}
+EOF
+}
+
+render_config_yaml() {
+  cat <<EOF
+# NetBird Enterprise server configuration.
+# Generated by getting-started-enterprise.sh. Mode 600.
+
+server:
+  listenAddress: ":80"
+  exposedAddress: "https://${NETBIRD_DOMAIN}:443"
+
+  metricsPort: 9090
+  healthcheckAddress: ":9000"
+
+  logLevel: "info"
+  logFile: "console"
+
+  # TLS is terminated by Caddy in front; leave this block empty.
+  tls:
+    certFile: ""
+    keyFile: ""
+    letsencrypt:
+      enabled: false
+
+  authSecret: "${NETBIRD_RELAY_AUTH_SECRET}"
+  dataDir: "/var/lib/netbird/"
+
+  disableAnonymousMetrics: false
+  disableGeoliteUpdate: false
+
+  auth:
+    issuer: "https://${NETBIRD_DOMAIN}/oauth2"
+    localAuthDisabled: false
+    signKeyRefreshEnabled: false
+    dashboardRedirectURIs:
+      - "https://${NETBIRD_DOMAIN}/nb-auth"
+      - "https://${NETBIRD_DOMAIN}/nb-silent-auth"
+    cliRedirectURIs:
+      - "http://localhost:53000/"
+
+  store:
+    engine: "postgres"
+    dsn: "${POSTGRES_DSN}"
+    encryptionKey: "${NETBIRD_ENCRYPTION_KEY}"
+
+  activityStore:
+    engine: "postgres"
+    dsn: "${POSTGRES_DSN}"
+EOF
+
+  if [[ "$NETBIRD_TRAFFIC_FLOW" == "yes" ]]; then
+    cat <<EOF
+
+  trafficFlow:
+    enabled: true
+    address: "https://${NETBIRD_DOMAIN}:443"
+    interval: "60s"
+EOF
+  fi
+}
+
+init_environment

infrastructure_files/migrate-to-enterprise.sh

@@ -0,0 +1,638 @@
+#!/bin/bash
+
+set -e
+set -o pipefail
+
+# NetBird — community combined → Enterprise combined migration
+#
+# Non-destructive migration: produces docker-compose.override.yml (auto-loaded
+# by docker compose) and config.yaml.enterprise alongside the operator's
+# existing files. Original docker-compose.yml and config.yaml are never
+# modified.
+#
+# Steps (all optional, asked interactively):
+#   1. Image swap         — replace community images with enterprise cloud images.
+#   2. Postgres migration — add Postgres, migrate SQLite data via migrate-store.
+#   3. Traffic flow       — add NATS + flow-enricher + flow-receiver.
+#
+# To revert:
+#   docker compose down
+#   rm -f docker-compose.override.yml config.yaml.enterprise
+#   # If Postgres migration was done, also restore the SQLite backup printed
+#   # at the end of this script's run.
+#   docker compose up -d
+
+OVERRIDE_FILE="docker-compose.override.yml"
+ENTERPRISE_CONFIG_FILE="config.yaml.enterprise"
+
+check_docker_compose() {
+  if command -v docker-compose &> /dev/null; then
+    echo "docker-compose"
+    return
+  fi
+  if docker compose --help &> /dev/null; then
+    echo "docker compose"
+    return
+  fi
+  echo "docker-compose is not installed or not in PATH." > /dev/stderr
+  exit 1
+}
+
+check_yq() {
+  if ! command -v yq &> /dev/null; then
+    cat > /dev/stderr <<'EOF'
+yq is required to parse and update YAML safely.
+
+  macOS:   brew install yq
+  Linux:   https://github.com/mikefarah/yq/releases (download binary into PATH)
+  Debian:  apt-get install yq   (Note: must be the mikefarah Go yq, not the Python wrapper.)
+
+EOF
+    exit 1
+  fi
+  if ! yq --version 2>&1 | grep -q "mikefarah"; then
+    echo "yq is present but appears to be the wrong implementation. The mikefarah Go-based yq is required (https://github.com/mikefarah/yq)." > /dev/stderr
+    exit 1
+  fi
+}
+
+check_openssl() {
+  if ! command -v openssl &> /dev/null; then
+    echo "openssl is not installed or not in PATH." > /dev/stderr
+    exit 1
+  fi
+}
+
+rand_password() {
+  openssl rand -hex 32
+}
+
+read_required() {
+  local prompt="$1"
+  local value=""
+  while [[ -z "$value" ]]; do
+    echo -n "$prompt: " > /dev/stderr
+    read -r value < /dev/tty
+    if [[ -z "$value" ]]; then
+      echo "Value cannot be empty." > /dev/stderr
+    fi
+  done
+  echo "$value"
+}
+
+read_secret() {
+  local prompt="$1"
+  local value=""
+  while [[ -z "$value" ]]; do
+    echo -n "$prompt: " > /dev/stderr
+    read -rs value < /dev/tty
+    echo "" > /dev/stderr
+    if [[ -z "$value" ]]; then
+      echo "Value cannot be empty." > /dev/stderr
+    fi
+  done
+  echo "$value"
+}
+
+read_yes_no() {
+  local prompt="$1"
+  local default="${2:-n}"
+  local hint
+  if [[ "$default" == "y" ]]; then
+    hint="[Y/n]"
+  else
+    hint="[y/N]"
+  fi
+  echo -n "${prompt} ${hint}: " > /dev/stderr
+  local ans=""
+  read -r ans < /dev/tty
+  if [[ -z "$ans" ]]; then
+    ans="$default"
+  fi
+  case "$ans" in
+    [Yy] | [Yy][Ee][Ss]) echo "yes" ;;
+    *) echo "no" ;;
+  esac
+}
+
+# ---------------------------------------------------------------------------
+# Detection — read the operator's existing compose to find service names and
+# paths we need to override. Bail loudly if shape isn't recognised.
+# ---------------------------------------------------------------------------
+
+detect_combined_service() {
+  yq eval '.services | to_entries | map(select(.value.image | test("^netbirdio/netbird-server"))) | .[0].key // ""' "$COMPOSE_FILE"
+}
+
+detect_dashboard_service() {
+  yq eval '.services | to_entries | map(select(.value.image | test("^netbirdio/dashboard"))) | .[0].key // ""' "$COMPOSE_FILE"
+}
+
+detect_config_yaml_host_path() {
+  yq eval ".services[\"$COMBINED_SERVICE\"].volumes[] | select(. | test(\":/etc/netbird/config.yaml\")) | sub(\":/etc/netbird/config.yaml.*\"; \"\") // \"\"" "$COMPOSE_FILE" | head -1
+}
+
+detect_data_volume() {
+  yq eval ".services[\"$COMBINED_SERVICE\"].volumes[] | select(. | test(\":/var/lib/netbird\")) | sub(\":/var/lib/netbird.*\"; \"\") // \"\"" "$COMPOSE_FILE" | head -1
+}
+
+detect_exposed_address() {
+  yq eval '.server.exposedAddress // ""' "$CONFIG_YAML_HOST"
+}
+
+detect_compose_network() {
+  local tag
+  tag=$(yq eval ".services[\"$COMBINED_SERVICE\"].networks | tag" "$COMPOSE_FILE" 2>/dev/null)
+  case "$tag" in
+    "!!seq")
+      yq eval ".services[\"$COMBINED_SERVICE\"].networks[0]" "$COMPOSE_FILE"
+      ;;
+    "!!map")
+      yq eval ".services[\"$COMBINED_SERVICE\"].networks | keys | .[0]" "$COMPOSE_FILE"
+      ;;
+    *)
+      echo "default"
+      ;;
+  esac
+}
+
+# ---------------------------------------------------------------------------
+# Renderers
+# ---------------------------------------------------------------------------
+
+# Build docker-compose.override.yml from the steps the operator selected.
+# Service names match what we detected on the operator's side.
+render_override() {
+  cat <<EOF
+# Generated by migrate-to-enterprise.sh. Mode 644.
+# Merged with docker-compose.yml automatically by Docker Compose.
+# Remove this file (and config.yaml.enterprise if present) to revert.
+
+services:
+  ${DASHBOARD_SERVICE}:
+    image: \${NETBIRD_DASHBOARD_IMAGE:-ghcr.io/netbirdio/dashboard-cloud:latest}
+
+  ${COMBINED_SERVICE}:
+    image: \${NETBIRD_SERVER_IMAGE:-ghcr.io/netbirdio/netbird-server-cloud:latest}
+    environment:
+      NB_LICENSE_KEY: \${NB_LICENSE_KEY}
+      NETBIRD_LICENSE_SERVER_BASE_URL: \${NETBIRD_LICENSE_SERVER_BASE_URL}
+EOF
+
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    cat <<EOF
+    depends_on:
+      postgres:
+        condition: service_healthy
+    volumes:
+      - ./${ENTERPRISE_CONFIG_FILE}:/etc/netbird/config.yaml.enterprise:ro
+    command: ["--config", "/etc/netbird/config.yaml.enterprise"]
+
+  postgres:
+    image: postgres:17
+    container_name: netbird-postgres
+    restart: unless-stopped
+    networks: [${COMPOSE_NETWORK}]
+    environment:
+      POSTGRES_USER: netbird
+      POSTGRES_PASSWORD: \${POSTGRES_PASSWORD}
+      POSTGRES_DB: netbird
+    volumes:
+      - netbird_postgres:/var/lib/postgresql/data
+    healthcheck:
+      test: ["CMD-SHELL", "pg_isready -U netbird -d netbird"]
+      interval: 5s
+      timeout: 5s
+      retries: 20
+EOF
+  fi
+
+  if [[ "$ENABLE_FLOW" == "yes" ]]; then
+    cat <<EOF
+
+  nats:
+    image: nats:2
+    container_name: netbird-nats
+    restart: unless-stopped
+    networks: [${COMPOSE_NETWORK}]
+    command: ["-m", "8222", "--jetstream", "--store_dir", "/data"]
+    volumes:
+      - netbird_nats_data:/data
+
+  flow-enricher:
+    image: ghcr.io/netbirdio/flow-enricher-cloud:latest
+    container_name: netbird-flow-enricher
+    restart: unless-stopped
+    networks: [${COMPOSE_NETWORK}]
+    depends_on:
+      postgres:
+        condition: service_healthy
+      nats:
+        condition: service_started
+    environment:
+      NB_LICENSE_KEY: \${NB_LICENSE_KEY}
+      NETBIRD_LICENSE_SERVER_BASE_URL: \${NETBIRD_LICENSE_SERVER_BASE_URL}
+      NB_DATADIR: /var/lib/netbird
+      NB_MANAGEMENT_STORE_ENGINE: postgres
+      NB_MANAGEMENT_POSTGRES_DSN: "host=postgres user=netbird password=\${POSTGRES_PASSWORD} dbname=netbird port=5432 sslmode=disable"
+      NB_STORE_ENGINE_POSTGRES_DSN: "host=postgres user=netbird password=\${POSTGRES_PASSWORD} dbname=netbird port=5432 sslmode=disable"
+      NB_TRAFFIC_EVENT_STORE_ENGINE: postgres
+      NB_TRAFFIC_EVENT_POSTGRES_DSN: "host=postgres user=netbird password=\${POSTGRES_PASSWORD} dbname=netbird port=5432 sslmode=disable"
+      NB_MANAGEMENT_STORE_KEY: \${NETBIRD_ENCRYPTION_KEY}
+      NB_FLOW_ADAPTER_TYPE: nats
+      NB_FLOW_NATS_ENDPOINTS: nats://nats:4222
+      NB_FLOW_NATS_STREAM: traffic-events
+      NB_METRICS_PORT: 9091
+      NB_PERSISTENCE_RETENTION_PERIOD: 168h
+
+  flow-receiver:
+    image: ghcr.io/netbirdio/flow-receiver-cloud:latest
+    container_name: netbird-flow-receiver
+    restart: unless-stopped
+    networks: [${COMPOSE_NETWORK}]
+    depends_on:
+      nats:
+        condition: service_started
+    environment:
+      NB_LICENSE_KEY: \${NB_LICENSE_KEY}
+      NETBIRD_LICENSE_SERVER_BASE_URL: \${NETBIRD_LICENSE_SERVER_BASE_URL}
+      NB_FLOW_LISTEN_PORT: 80
+      NB_FLOW_ADAPTER_TYPE: nats
+      NB_FLOW_NATS_ENDPOINTS: nats://nats:4222
+      NB_FLOW_NATS_STREAM: traffic-events
+      NB_FLOW_AUTH_SECRET: \${NB_FLOW_AUTH_SECRET}
+    labels:
+      - traefik.enable=true
+      - traefik.http.routers.netbird-flow.rule=Host(\`${NETBIRD_HOSTNAME}\`) && PathPrefix(\`/flow.FlowService/\`)
+      - traefik.http.routers.netbird-flow.entrypoints=websecure
+      - traefik.http.routers.netbird-flow.tls=true
+      - traefik.http.routers.netbird-flow.tls.certresolver=letsencrypt
+      - traefik.http.routers.netbird-flow.service=netbird-flow-h2c
+      - traefik.http.routers.netbird-flow.priority=100
+      - traefik.http.services.netbird-flow-h2c.loadbalancer.server.port=80
+      - traefik.http.services.netbird-flow-h2c.loadbalancer.server.scheme=h2c
+EOF
+  fi
+
+  # Volume declarations for anything new the override introduced
+  local has_volumes="no"
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]] || [[ "$ENABLE_FLOW" == "yes" ]]; then
+    has_volumes="yes"
+  fi
+
+  if [[ "$has_volumes" == "yes" ]]; then
+    cat <<EOF
+
+volumes:
+EOF
+    if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+      echo "  netbird_postgres:"
+    fi
+    if [[ "$ENABLE_FLOW" == "yes" ]]; then
+      echo "  netbird_nats_data:"
+    fi
+  fi
+}
+
+# Build config.yaml.enterprise by yq-editing the operator's existing
+# config.yaml. We don't touch the original file.
+render_enterprise_config() {
+  local pg_dsn="host=postgres user=netbird password=${POSTGRES_PASSWORD} dbname=netbird port=5432 sslmode=disable"
+
+  yq eval "
+    .server.store.engine = \"postgres\" |
+    .server.store.dsn = \"$pg_dsn\" |
+    .server.activityStore.engine = \"postgres\" |
+    .server.activityStore.dsn = \"$pg_dsn\" |
+    .server.authStore.engine = \"postgres\" |
+    .server.authStore.dsn = \"$pg_dsn\"
+  " "$CONFIG_YAML_HOST" > "$ENTERPRISE_CONFIG_FILE"
+
+  if [[ "$ENABLE_FLOW" == "yes" ]]; then
+    local flow_addr="${NETBIRD_DOMAIN}"
+    yq eval -i "
+      .server.trafficFlow.enabled = true |
+      .server.trafficFlow.address = \"$flow_addr\" |
+      .server.trafficFlow.interval = \"60s\"
+    " "$ENTERPRISE_CONFIG_FILE"
+  fi
+}
+
+# ---------------------------------------------------------------------------
+# Execution steps
+# ---------------------------------------------------------------------------
+
+resolve_data_volume() {
+  local short="$1"
+  local actual
+  # Resolve project-prefixed volume name from Docker Compose config first.
+  actual=$($DOCKER_COMPOSE_COMMAND config 2>/dev/null | yq eval ".volumes.\"$short\".name" - 2>/dev/null)
+  if [[ -n "$actual" && "$actual" != "null" ]]; then
+    echo "$actual"
+    return
+  fi
+  # Relative bind mount: docker-compose resolves it against the compose
+  # file's directory, but `docker run -v` resolves it against the current
+  # working directory. Normalize to an absolute path so both interpretations
+  # agree (and the printed revert command works from any CWD).
+  if [[ "$short" == ./* || "$short" == ../* ]]; then
+    local compose_dir
+    compose_dir="$(cd "$(dirname "$COMPOSE_FILE")" && pwd)"
+    (
+      cd "$compose_dir"
+      cd "$(dirname "$short")"
+      printf '%s/%s\n' "$(pwd)" "$(basename "$short")"
+    )
+    return
+  fi
+  # Not a named volume (e.g. an absolute bind-mount path) — use it as-is.
+  echo "$short"
+}
+
+backup_sqlite() {
+  BACKUP_DIR="$(pwd)/backups/sqlite-pre-enterprise-$(date +%Y%m%d-%H%M%S)"
+  mkdir -p "$BACKUP_DIR"
+  local data_volume_actual
+  data_volume_actual=$(resolve_data_volume "$DATA_VOLUME")
+  echo "Backing up SQLite store from volume '$data_volume_actual' to $BACKUP_DIR ..."
+  docker run --rm \
+    -v "${data_volume_actual}:/var/lib/netbird:ro" \
+    -v "${BACKUP_DIR}:/backup" \
+    busybox \
+    sh -c 'cp -a /var/lib/netbird/. /backup/ 2>/dev/null || true'
+  local copied
+  copied=$(find "$BACKUP_DIR" -mindepth 1 | head -1)
+  if [[ -z "$copied" ]]; then
+    echo "  ⚠ Backup directory is empty — the volume '$data_volume_actual' didn't contain data. Aborting." > /dev/stderr
+    exit 1
+  fi
+  echo "  done"
+}
+
+run_migrate_store() {
+  echo "Running migrate-store (SQLite → Postgres) ..."
+  $DOCKER_COMPOSE_COMMAND run --rm "$COMBINED_SERVICE" migrate-store --config /etc/netbird/config.yaml.enterprise --verify
+  echo "  done"
+}
+
+# ---------------------------------------------------------------------------
+# Main
+# ---------------------------------------------------------------------------
+
+init_migration() {
+  DOCKER_COMPOSE_COMMAND=$(check_docker_compose)
+  check_yq
+  check_openssl
+
+  COMPOSE_FILE="${COMPOSE_FILE:-docker-compose.yml}"
+
+  if [[ ! -f "$COMPOSE_FILE" ]]; then
+    echo "$COMPOSE_FILE not found in $(pwd)." > /dev/stderr
+    exit 1
+  fi
+  if [[ -f "$OVERRIDE_FILE" ]] || [[ -f "$ENTERPRISE_CONFIG_FILE" ]]; then
+    echo "Migration artifacts already exist in $(pwd):"
+    [[ -f "$OVERRIDE_FILE" ]] && echo "  $OVERRIDE_FILE"
+    [[ -f "$ENTERPRISE_CONFIG_FILE" ]] && echo "  $ENTERPRISE_CONFIG_FILE"
+    echo ""
+    echo "Either you've already migrated, or a previous run was interrupted."
+    echo "To re-run cleanly: rm -f $OVERRIDE_FILE $ENTERPRISE_CONFIG_FILE"
+    exit 1
+  fi
+
+  COMBINED_SERVICE=$(detect_combined_service)
+  DASHBOARD_SERVICE=$(detect_dashboard_service)
+  CONFIG_YAML_HOST=$(detect_config_yaml_host_path)
+  DATA_VOLUME=$(detect_data_volume)
+  COMPOSE_NETWORK=$(detect_compose_network)
+
+  if [[ -z "$COMBINED_SERVICE" ]]; then
+    echo "Could not find a service running netbirdio/netbird-server* in $COMPOSE_FILE." > /dev/stderr
+    echo "This script targets the community combined-server deployment." > /dev/stderr
+    exit 1
+  fi
+  if [[ -z "$DASHBOARD_SERVICE" ]]; then
+    echo "Could not find a service running netbirdio/dashboard* in $COMPOSE_FILE." > /dev/stderr
+    exit 1
+  fi
+  if [[ -z "$CONFIG_YAML_HOST" ]]; then
+    echo "Could not find a config.yaml mount on $COMBINED_SERVICE (expected to bind-mount to /etc/netbird/config.yaml)." > /dev/stderr
+    exit 1
+  fi
+  if [[ ! -f "$CONFIG_YAML_HOST" ]]; then
+    echo "config.yaml host file not found at $CONFIG_YAML_HOST." > /dev/stderr
+    exit 1
+  fi
+  if [[ -z "$DATA_VOLUME" ]]; then
+    echo "Could not find a volume mounted at /var/lib/netbird on $COMBINED_SERVICE." > /dev/stderr
+    exit 1
+  fi
+
+  echo "Detected existing deployment:"
+  echo "  Combined service: $COMBINED_SERVICE"
+  echo "  Dashboard:        $DASHBOARD_SERVICE"
+  echo "  config.yaml:      $CONFIG_YAML_HOST"
+  echo "  Data volume:      $DATA_VOLUME"
+  echo "  Network:          $COMPOSE_NETWORK"
+  echo ""
+
+  local proceed
+  proceed=$(read_yes_no "Proceed with migration?" "y")
+  if [[ "$proceed" != "yes" ]]; then
+    echo "Aborted."
+    exit 0
+  fi
+
+  # Step 1 — always (this is the point of the script)
+  MIGRATE_IMAGES="yes"
+  echo ""
+  echo "Step 1: Image swap (community → Enterprise). License key required."
+  NB_LICENSE_KEY=$(read_secret "  License key")
+  GHCR_USERNAME="netbirdExtAccess1"
+  GHCR_TOKEN=$(read_secret "  GHCR token (input hidden)")
+
+  # Step 2 — optional
+  echo ""
+  MIGRATE_POSTGRES=$(read_yes_no "Step 2: Migrate storage from SQLite to Postgres? (recommended)" "n")
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    echo ""
+    echo "  ⚠  Data will be migrated from SQLite to Postgres. The SQLite store"
+    echo "     will be backed up automatically. To fully revert later, restore"
+    echo "     that backup and delete docker-compose.override.yml +"
+    echo "     config.yaml.enterprise."
+    local confirm
+    confirm=$(read_yes_no "  Continue?" "y")
+    if [[ "$confirm" != "yes" ]]; then
+      MIGRATE_POSTGRES="no"
+      echo "  Skipping Postgres migration."
+    else
+      POSTGRES_PASSWORD=$(rand_password)
+    fi
+  fi
+
+  # Step 3 — optional, only if Postgres is on (flow requires Postgres)
+  echo ""
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    ENABLE_FLOW=$(read_yes_no "Step 3: Enable traffic flow? (requires Postgres)" "n")
+    if [[ "$ENABLE_FLOW" == "yes" ]]; then
+      # Auth secret MUST match server.authSecret from config.yaml
+      NB_FLOW_AUTH_SECRET=$(yq eval '.server.authSecret // ""' "$CONFIG_YAML_HOST")
+      if [[ -z "$NB_FLOW_AUTH_SECRET" ]] || [[ "$NB_FLOW_AUTH_SECRET" == "null" ]]; then
+        echo "Could not read server.authSecret from $CONFIG_YAML_HOST." > /dev/stderr
+        echo "Flow receiver auth must match the combined server's authSecret." > /dev/stderr
+        exit 1
+      fi
+
+      NETBIRD_DOMAIN=$(detect_exposed_address)
+      if [[ -z "$NETBIRD_DOMAIN" ]] || [[ "$NETBIRD_DOMAIN" == "null" ]]; then
+        NETBIRD_DOMAIN=$(read_required "  Public NetBird URL (e.g. https://netbird.example.com)")
+      fi
+      # Strip protocol + port to leave just the hostname for the Traefik Host() rule.
+      NETBIRD_HOSTNAME=$(echo "$NETBIRD_DOMAIN" | sed -E 's,^https?://,,' | sed 's,:.*,,' | sed 's,/.*,,')
+
+      # We need the encryption key from the existing config.yaml for the enricher
+      NETBIRD_ENCRYPTION_KEY=$(yq eval '.server.store.encryptionKey // ""' "$CONFIG_YAML_HOST")
+      if [[ -z "$NETBIRD_ENCRYPTION_KEY" ]] || [[ "$NETBIRD_ENCRYPTION_KEY" == "null" ]]; then
+        echo "Could not read server.store.encryptionKey from $CONFIG_YAML_HOST." > /dev/stderr
+        exit 1
+      fi
+    fi
+  else
+    ENABLE_FLOW="no"
+    echo "Step 3 (traffic flow) skipped — requires Postgres."
+  fi
+}
+
+apply_changes() {
+  echo ""
+  echo "Writing $OVERRIDE_FILE ..."
+  install -m 644 /dev/null "$OVERRIDE_FILE"
+  render_override > "$OVERRIDE_FILE"
+
+  if [[ -z "${NETBIRD_LICENSE_SERVER_BASE_URL:-}" ]]; then
+    sed -i.bak '/NETBIRD_LICENSE_SERVER_BASE_URL/d' "$OVERRIDE_FILE" && rm -f "$OVERRIDE_FILE.bak"
+  fi
+
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    echo "Writing $ENTERPRISE_CONFIG_FILE ..."
+    install -m 600 /dev/null "$ENTERPRISE_CONFIG_FILE"
+    render_enterprise_config
+  fi
+
+  # Persist secrets that the override file references via env interpolation.
+  # We write them to a .env file in the current directory; docker compose
+  # picks it up automatically.
+  echo "Writing .env additions (mode 600) ..."
+  local ENV_FILE=".env"
+  touch "$ENV_FILE"
+  chmod 600 "$ENV_FILE"
+  {
+    echo ""
+    echo "# Added by migrate-to-enterprise.sh on $(date -u +%Y-%m-%dT%H:%M:%SZ)"
+    echo "NB_LICENSE_KEY=${NB_LICENSE_KEY}"
+    if [[ -n "${NETBIRD_LICENSE_SERVER_BASE_URL:-}" ]]; then
+      echo "NETBIRD_LICENSE_SERVER_BASE_URL=${NETBIRD_LICENSE_SERVER_BASE_URL}"
+    fi
+    if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+      echo "POSTGRES_PASSWORD=${POSTGRES_PASSWORD}"
+    fi
+    if [[ "$ENABLE_FLOW" == "yes" ]]; then
+      echo "NB_FLOW_AUTH_SECRET=${NB_FLOW_AUTH_SECRET}"
+      echo "NETBIRD_ENCRYPTION_KEY=${NETBIRD_ENCRYPTION_KEY}"
+    fi
+  } >> "$ENV_FILE"
+
+  echo ""
+  echo "Logging in to ghcr.io ..."
+  printf '%s' "$GHCR_TOKEN" | docker login ghcr.io -u "$GHCR_USERNAME" --password-stdin
+  unset GHCR_TOKEN
+
+  echo ""
+  echo "Pulling enterprise images ..."
+  $DOCKER_COMPOSE_COMMAND pull
+
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    echo ""
+    echo "Stopping existing services (volumes preserved) ..."
+    $DOCKER_COMPOSE_COMMAND down
+
+    backup_sqlite
+
+    echo ""
+    echo "Starting Postgres ..."
+    $DOCKER_COMPOSE_COMMAND up -d postgres
+
+    # Wait for healthy
+    local counter=0
+    echo -n "Waiting for Postgres to become ready"
+    while ! $DOCKER_COMPOSE_COMMAND exec -T postgres pg_isready -U netbird -d netbird &> /dev/null; do
+      echo -n " ."
+      sleep 2
+      counter=$((counter + 1))
+      if [[ $counter -ge 60 ]]; then
+        echo ""
+        echo "Postgres did not become ready in 120s. Recent logs:"
+        $DOCKER_COMPOSE_COMMAND logs --tail=20 postgres
+        exit 1
+      fi
+    done
+    echo " done"
+
+    run_migrate_store
+  fi
+
+  echo ""
+  echo "Bringing up all services ..."
+  $DOCKER_COMPOSE_COMMAND up -d
+
+  echo ""
+  echo "Migration complete."
+}
+
+print_summary() {
+  echo ""
+  echo "──────────────────────────────────────────────────────────────────────"
+  echo " Summary"
+  echo "──────────────────────────────────────────────────────────────────────"
+  echo "  Images:           swapped to enterprise"
+  [[ "$MIGRATE_POSTGRES" == "yes" ]] && echo "  Storage:          Postgres (data migrated from SQLite)"
+  [[ "$MIGRATE_POSTGRES" != "yes" ]] && echo "  Storage:          SQLite (unchanged)"
+  [[ "$ENABLE_FLOW" == "yes" ]] && echo "  Traffic flow:     enabled"
+  [[ "$ENABLE_FLOW" != "yes" ]] && echo "  Traffic flow:     disabled"
+  echo ""
+  echo "  Generated files (next to your docker-compose.yml):"
+  echo "    $OVERRIDE_FILE"
+  [[ "$MIGRATE_POSTGRES" == "yes" ]] && echo "    $ENTERPRISE_CONFIG_FILE"
+  echo "    .env  (license key + secrets, mode 600)"
+  [[ "$MIGRATE_POSTGRES" == "yes" ]] && echo "    backups/sqlite-pre-enterprise-*/  (SQLite backup)"
+  echo ""
+  echo " Tail logs:"
+  echo "   $DOCKER_COMPOSE_COMMAND logs -f $COMBINED_SERVICE"
+  echo ""
+  echo "──────────────────────────────────────────────────────────────────────"
+  echo " To revert"
+  echo "──────────────────────────────────────────────────────────────────────"
+  echo "  $DOCKER_COMPOSE_COMMAND down"
+  if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    # Resolve project-prefixed volume names now (before override is removed).
+    local pg_volume data_volume_actual
+    pg_volume=$(resolve_data_volume "netbird_postgres")
+    data_volume_actual=$(resolve_data_volume "$DATA_VOLUME")
+    echo "  # Remove the Postgres volume FIRST, before deleting the override file:"
+    echo "  docker volume rm $pg_volume"
+    echo "  # Restore SQLite from the backup created during this run:"
+    echo "  docker run --rm -v ${data_volume_actual}:/var/lib/netbird -v ${BACKUP_DIR}:/backup busybox sh -c 'cp -a /backup/. /var/lib/netbird/'"
+  fi
+  echo "  rm -f $OVERRIDE_FILE $ENTERPRISE_CONFIG_FILE"
+  echo "  # Remove migrate-to-enterprise.sh additions from .env (search for the timestamp marker)"
+  echo "  $DOCKER_COMPOSE_COMMAND up -d"
+  echo "──────────────────────────────────────────────────────────────────────"
+}
+
+# ---------------------------------------------------------------------------
+# Run
+# ---------------------------------------------------------------------------
+
+init_migration
+apply_changes
+print_summary

management/internals/controllers/network_map/controller/controller.go

@@ -610,12 +610,10 @@ func (c *Controller) GetValidatedPeerWithMap(ctx context.Context, isRequiresAppr
 		return nil, nil, 0, err
 	}
 
-	startPosture := time.Now()
 	postureChecks, err := c.getPeerPostureChecks(account, peerID)
 	if err != nil {
 		return nil, nil, 0, err
 	}
-	log.WithContext(ctx).Debugf("getPeerPostureChecks took %s", time.Since(startPosture))
 
 	accountZones, err := c.repo.GetAccountZones(ctx, account.Id)
 	if err != nil {

management/internals/shared/grpc/loginfilter.go

@@ -11,7 +11,7 @@ import (
 
 const (
 	reconnThreshold   = 5 * time.Minute
-	baseBlockDuration = 10 * time.Minute // Duration for which a peer is banned after exceeding the reconnection limit
+	baseBlockDuration = 30 * time.Minute // Duration for which a peer is banned after exceeding the reconnection limit
 	reconnLimitForBan = 30               // Number of reconnections within the reconnTreshold that triggers a ban
 	metaChangeLimit   = 3                // Number of reconnections with different metadata that triggers a ban of one peer
 )
@@ -139,22 +139,13 @@ func (l *loginFilter) addLogin(wgPubKey string, metaHash uint64) {
 	state.lastSeen = now
 }
 
-func metaHash(meta nbpeer.PeerSystemMeta, pubip string) uint64 {
+func metaHash(meta nbpeer.PeerSystemMeta) uint64 {
 	h := fnv.New64a()
 
-	h.Write([]byte(meta.WtVersion))
 	h.Write([]byte(meta.OSVersion))
 	h.Write([]byte(meta.KernelVersion))
 	h.Write([]byte(meta.Hostname))
 	h.Write([]byte(meta.SystemSerialNumber))
-	h.Write([]byte(pubip))
 
-	macs := uint64(0)
-	for _, na := range meta.NetworkAddresses {
-		for _, r := range na.Mac {
-			macs += uint64(r)
-		}
-	}
-
-	return h.Sum64() + macs
+	return h.Sum64()
 }

management/internals/shared/grpc/loginfilter_test.go

@@ -164,9 +164,7 @@ func BenchmarkHashingMethods(b *testing.B) {
 		KernelVersion:      "5.15.0-76-generic",
 		Hostname:           "prod-server-database-01",
 		SystemSerialNumber: "PC-1234567890",
-		NetworkAddresses:   []nbpeer.NetworkAddress{{Mac: "00:1B:44:11:3A:B7"}, {Mac: "00:1B:44:11:3A:B8"}},
 	}
-	pubip := "8.8.8.8"
 
 	var resultString string
 	var resultUint uint64
@@ -175,7 +173,7 @@ func BenchmarkHashingMethods(b *testing.B) {
 		b.ReportAllocs()
 		b.ResetTimer()
 		for i := 0; i < b.N; i++ {
-			resultString = builderString(meta, pubip)
+			resultString = builderString(meta)
 		}
 	})
 
@@ -183,7 +181,7 @@ func BenchmarkHashingMethods(b *testing.B) {
 		b.ReportAllocs()
 		b.ResetTimer()
 		for i := 0; i < b.N; i++ {
-			resultString = fnvHashToString(meta, pubip)
+			resultString = fnvHashToString(meta)
 		}
 	})
 
@@ -191,7 +189,7 @@ func BenchmarkHashingMethods(b *testing.B) {
 		b.ReportAllocs()
 		b.ResetTimer()
 		for i := 0; i < b.N; i++ {
-			resultUint = metaHash(meta, pubip)
+			resultUint = metaHash(meta)
 		}
 	})
 
@@ -199,29 +197,20 @@ func BenchmarkHashingMethods(b *testing.B) {
 	_ = resultUint
 }
 
-func fnvHashToString(meta nbpeer.PeerSystemMeta, pubip string) string {
+func fnvHashToString(meta nbpeer.PeerSystemMeta) string {
 	h := fnv.New64a()
 
-	if len(meta.NetworkAddresses) != 0 {
-		for _, na := range meta.NetworkAddresses {
-			h.Write([]byte(na.Mac))
-		}
-	}
-
 	h.Write([]byte(meta.WtVersion))
 	h.Write([]byte(meta.OSVersion))
 	h.Write([]byte(meta.KernelVersion))
 	h.Write([]byte(meta.Hostname))
 	h.Write([]byte(meta.SystemSerialNumber))
-	h.Write([]byte(pubip))
 
 	return strconv.FormatUint(h.Sum64(), 16)
 }
 
-func builderString(meta nbpeer.PeerSystemMeta, pubip string) string {
-	mac := getMacAddress(meta.NetworkAddresses)
-	estimatedSize := len(meta.WtVersion) + len(meta.OSVersion) + len(meta.KernelVersion) + len(meta.Hostname) + len(meta.SystemSerialNumber) +
-		len(pubip) + len(mac) + 6
+func builderString(meta nbpeer.PeerSystemMeta) string {
+	estimatedSize := len(meta.WtVersion) + len(meta.OSVersion) + len(meta.KernelVersion) + len(meta.Hostname) + len(meta.SystemSerialNumber) + 4
 
 	var b strings.Builder
 	b.Grow(estimatedSize)
@@ -235,23 +224,10 @@ func builderString(meta nbpeer.PeerSystemMeta, pubip string) string {
 	b.WriteString(meta.Hostname)
 	b.WriteByte('|')
 	b.WriteString(meta.SystemSerialNumber)
-	b.WriteByte('|')
-	b.WriteString(pubip)
 
 	return b.String()
 }
 
-func getMacAddress(nas []nbpeer.NetworkAddress) string {
-	if len(nas) == 0 {
-		return ""
-	}
-	macs := make([]string, 0, len(nas))
-	for _, na := range nas {
-		macs = append(macs, na.Mac)
-	}
-	return strings.Join(macs, "/")
-}
-
 func BenchmarkLoginFilter_ParallelLoad(b *testing.B) {
 	filter := newLoginFilterWithCfg(testAdvancedCfg())
 	numKeys := 100000

management/internals/shared/grpc/server.go

@@ -254,7 +254,7 @@ func (s *Server) Sync(req *proto.EncryptedMessage, srv proto.ManagementService_S
 		return mapError(ctx, err)
 	}
 
-	metahashed := metaHash(peerMeta, sRealIP)
+	metahashed := metaHash(peerMeta)
 	if userID == "" && !s.loginFilter.allowLogin(peerKey.String(), metahashed) {
 		if s.appMetrics != nil {
 			s.appMetrics.GRPCMetrics().CountSyncRequestBlocked()
@@ -306,7 +306,7 @@ func (s *Server) Sync(req *proto.EncryptedMessage, srv proto.ManagementService_S
 		log.WithContext(ctx).Tracef("peer system meta has to be provided on sync. Peer %s, remote addr %s", peerKey.String(), realIP)
 	}
 
-	metahash := metaHash(peerMeta, realIP.String())
+	metahash := metaHash(peerMeta)
 	s.loginFilter.addLogin(peerKey.String(), metahash)
 
 	peer, netMap, postureChecks, dnsFwdPort, err := s.accountManager.SyncAndMarkPeer(ctx, accountID, peerKey.String(), peerMeta, realIP, syncStart)
@@ -732,7 +732,7 @@ func (s *Server) Login(ctx context.Context, req *proto.EncryptedMessage) (*proto
 	}
 
 	peerMeta := extractPeerMeta(ctx, loginReq.GetMeta())
-	metahashed := metaHash(peerMeta, sRealIP)
+	metahashed := metaHash(peerMeta)
 	if !s.loginFilter.allowLogin(peerKey.String(), metahashed) {
 		if s.logBlockedPeers {
 			log.WithContext(ctx).Tracef("peer %s with meta hash %d is blocked from login", peerKey.String(), metahashed)
@@ -788,7 +788,11 @@ func (s *Server) Login(ctx context.Context, req *proto.EncryptedMessage) (*proto
 		ExtraDNSLabels:  loginReq.GetDnsLabels(),
 	})
 	if err != nil {
-		log.WithContext(ctx).Warnf("failed logging in peer %s: %s", peerKey, err)
+		if errors.Is(err, internalStatus.ErrNoAuthMethodProvided) {
+			log.WithContext(ctx).Tracef("failed logging in peer %s: %s", peerKey, err)
+		} else {
+			log.WithContext(ctx).Warnf("failed logging in peer %s: %s", peerKey, err)
+		}
 		return nil, mapError(ctx, err)
 	}
 
@@ -1205,7 +1209,7 @@ func (s *Server) SyncMeta(ctx context.Context, req *proto.EncryptedMessage) (*pr
 		return nil, msg
 	}
 
-	err = s.accountManager.SyncPeerMeta(ctx, peerKey.String(), extractPeerMeta(ctx, syncMetaReq.GetMeta()))
+	err = s.accountManager.SyncPeerMeta(ctx, peerKey.String(), extractPeerMeta(ctx, syncMetaReq.GetMeta()), realIP)
 	if err != nil {
 		return nil, mapError(ctx, err)
 	}
@@ -1254,7 +1258,10 @@ func (s *Server) Logout(ctx context.Context, req *proto.EncryptedMessage) (*prot
 func toProtocolChecks(ctx context.Context, postureChecks []*posture.Checks) []*proto.Checks {
 	protoChecks := make([]*proto.Checks, 0, len(postureChecks))
 	for _, postureCheck := range postureChecks {
-		protoChecks = append(protoChecks, toProtocolCheck(postureCheck))
+		check := toProtocolCheck(postureCheck)
+		if check != nil {
+			protoChecks = append(protoChecks, check)
+		}
 	}
 
 	return protoChecks
@@ -1278,5 +1285,9 @@ func toProtocolCheck(postureCheck *posture.Checks) *proto.Checks {
 		}
 	}
 
+	if len(protoCheck.Files) == 0 {
+		return nil
+	}
+
 	return protoCheck
 }

management/server/account.go

@@ -1889,12 +1889,12 @@ func domainIsUpToDate(domain string, domainCategory string, userAuth auth.UserAu
 // concurrent stream that started earlier loses the optimistic-lock race
 // in MarkPeerConnected and bails without writing.
 func (am *DefaultAccountManager) SyncAndMarkPeer(ctx context.Context, accountID string, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP, syncTime time.Time) (*nbpeer.Peer, *types.NetworkMap, []*posture.Checks, int64, error) {
-	peer, netMap, postureChecks, dnsfwdPort, err := am.SyncPeer(ctx, types.PeerSync{WireGuardPubKey: peerPubKey, Meta: meta}, accountID)
+	peer, netMap, postureChecks, dnsfwdPort, err := am.SyncPeer(ctx, types.PeerSync{WireGuardPubKey: peerPubKey, Meta: meta, RealIP: realIP}, accountID)
 	if err != nil {
 		return nil, nil, nil, 0, fmt.Errorf("error syncing peer: %w", err)
 	}
 
-	if err := am.MarkPeerConnected(ctx, peerPubKey, realIP, accountID, syncTime.UnixNano(), netMap); err != nil {
+	if err := am.MarkPeerConnected(ctx, peerPubKey, accountID, syncTime.UnixNano(), netMap); err != nil {
 		log.WithContext(ctx).Warnf("failed marking peer as connected %s %v", peerPubKey, err)
 	}
 
@@ -1914,13 +1914,13 @@ func (am *DefaultAccountManager) OnPeerDisconnected(ctx context.Context, account
 	return nil
 }
 
-func (am *DefaultAccountManager) SyncPeerMeta(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta) error {
+func (am *DefaultAccountManager) SyncPeerMeta(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP) error {
 	accountID, err := am.Store.GetAccountIDByPeerPubKey(ctx, peerPubKey)
 	if err != nil {
 		return err
 	}
 
-	_, _, _, _, err = am.SyncPeer(ctx, types.PeerSync{WireGuardPubKey: peerPubKey, Meta: meta, UpdateAccountPeers: true}, accountID)
+	_, _, _, _, err = am.SyncPeer(ctx, types.PeerSync{WireGuardPubKey: peerPubKey, Meta: meta, RealIP: realIP, UpdateAccountPeers: true}, accountID)
 	if err != nil {
 		return err
 	}

management/server/account/manager.go

@@ -62,7 +62,7 @@ type Manager interface {
 	GetUserFromUserAuth(ctx context.Context, userAuth auth.UserAuth) (*types.User, error)
 	ListUsers(ctx context.Context, accountID string) ([]*types.User, error)
 	GetPeers(ctx context.Context, accountID, userID, nameFilter, ipFilter string) ([]*nbpeer.Peer, error)
-	MarkPeerConnected(ctx context.Context, peerKey string, realIP net.IP, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error
+	MarkPeerConnected(ctx context.Context, peerKey string, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error
 	MarkPeerDisconnected(ctx context.Context, peerKey string, accountID string, sessionStartedAt int64) error
 	DeletePeer(ctx context.Context, accountID, peerID, userID string) error
 	UpdatePeer(ctx context.Context, accountID, userID string, p *nbpeer.Peer) (*nbpeer.Peer, error)
@@ -123,7 +123,7 @@ type Manager interface {
 	GetValidatedPeers(ctx context.Context, accountID string) (map[string]struct{}, map[string]string, error)
 	SyncAndMarkPeer(ctx context.Context, accountID string, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP, syncTime time.Time) (*nbpeer.Peer, *types.NetworkMap, []*posture.Checks, int64, error)
 	OnPeerDisconnected(ctx context.Context, accountID string, peerPubKey string, streamStartTime time.Time) error
-	SyncPeerMeta(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta) error
+	SyncPeerMeta(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP) error
 	FindExistingPostureCheck(accountID string, checks *posture.ChecksDefinition) (*posture.Checks, error)
 	GetAccountIDForPeerKey(ctx context.Context, peerKey string) (string, error)
 	GetAccountSettings(ctx context.Context, accountID string, userID string) (*types.Settings, error)

management/server/account/manager_mock.go

@@ -1323,17 +1323,17 @@ func (mr *MockManagerMockRecorder) ExtendPeerSession(ctx, peerPubKey, userID int
 }
 
 // MarkPeerConnected mocks base method.
-func (m *MockManager) MarkPeerConnected(ctx context.Context, peerKey string, realIP net.IP, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error {
+func (m *MockManager) MarkPeerConnected(ctx context.Context, peerKey string, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error {
 	m.ctrl.T.Helper()
-	ret := m.ctrl.Call(m, "MarkPeerConnected", ctx, peerKey, realIP, accountID, sessionStartedAt, nmap)
+	ret := m.ctrl.Call(m, "MarkPeerConnected", ctx, peerKey, accountID, sessionStartedAt, nmap)
 	ret0, _ := ret[0].(error)
 	return ret0
 }
 
 // MarkPeerConnected indicates an expected call of MarkPeerConnected.
-func (mr *MockManagerMockRecorder) MarkPeerConnected(ctx, peerKey, realIP, accountID, sessionStartedAt, nmap interface{}) *gomock.Call {
+func (mr *MockManagerMockRecorder) MarkPeerConnected(ctx, peerKey, accountID, sessionStartedAt, nmap interface{}) *gomock.Call {
 	mr.mock.ctrl.T.Helper()
-	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "MarkPeerConnected", reflect.TypeOf((*MockManager)(nil).MarkPeerConnected), ctx, peerKey, realIP, accountID, sessionStartedAt, nmap)
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "MarkPeerConnected", reflect.TypeOf((*MockManager)(nil).MarkPeerConnected), ctx, peerKey, accountID, sessionStartedAt, nmap)
 }
 
 // MarkPeerDisconnected mocks base method.
@@ -1586,17 +1586,17 @@ func (mr *MockManagerMockRecorder) SyncPeer(ctx, sync, accountID interface{}) *g
 }
 
 // SyncPeerMeta mocks base method.
-func (m *MockManager) SyncPeerMeta(ctx context.Context, peerPubKey string, meta peer.PeerSystemMeta) error {
+func (m *MockManager) SyncPeerMeta(ctx context.Context, peerPubKey string, meta peer.PeerSystemMeta, realIP net.IP) error {
 	m.ctrl.T.Helper()
-	ret := m.ctrl.Call(m, "SyncPeerMeta", ctx, peerPubKey, meta)
+	ret := m.ctrl.Call(m, "SyncPeerMeta", ctx, peerPubKey, meta, realIP)
 	ret0, _ := ret[0].(error)
 	return ret0
 }
 
 // SyncPeerMeta indicates an expected call of SyncPeerMeta.
-func (mr *MockManagerMockRecorder) SyncPeerMeta(ctx, peerPubKey, meta interface{}) *gomock.Call {
+func (mr *MockManagerMockRecorder) SyncPeerMeta(ctx, peerPubKey, meta, realIP interface{}) *gomock.Call {
 	mr.mock.ctrl.T.Helper()
-	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "SyncPeerMeta", reflect.TypeOf((*MockManager)(nil).SyncPeerMeta), ctx, peerPubKey, meta)
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "SyncPeerMeta", reflect.TypeOf((*MockManager)(nil).SyncPeerMeta), ctx, peerPubKey, meta, realIP)
 }
 
 // SyncUserJWTGroups mocks base method.

management/server/account_test.go

@@ -1836,7 +1836,7 @@ func TestDefaultAccountManager_UpdatePeer_PeerLoginExpiration(t *testing.T) {
 	accountID, err := manager.GetAccountIDByUserID(context.Background(), auth.UserAuth{UserId: userID})
 	require.NoError(t, err, "unable to get the account")
 
-	err = manager.MarkPeerConnected(context.Background(), key.PublicKey().String(), nil, accountID, time.Now().UTC().UnixNano(), nil)
+	err = manager.MarkPeerConnected(context.Background(), key.PublicKey().String(), accountID, time.Now().UTC().UnixNano(), nil)
 	require.NoError(t, err, "unable to mark peer connected")
 
 	_, err = manager.UpdateAccountSettings(context.Background(), accountID, userID, &types.Settings{
@@ -1907,7 +1907,7 @@ func TestDefaultAccountManager_MarkPeerConnected_PeerLoginExpiration(t *testing.
 	require.NoError(t, err, "unable to get the account")
 
 	// when we mark peer as connected, the peer login expiration routine should trigger
-	err = manager.MarkPeerConnected(context.Background(), key.PublicKey().String(), nil, accountID, time.Now().UTC().UnixNano(), nil)
+	err = manager.MarkPeerConnected(context.Background(), key.PublicKey().String(), accountID, time.Now().UTC().UnixNano(), nil)
 	require.NoError(t, err, "unable to mark peer connected")
 
 	failed := waitTimeout(wg, time.Second)
@@ -1916,6 +1916,117 @@ func TestDefaultAccountManager_MarkPeerConnected_PeerLoginExpiration(t *testing.
 	}
 }
 
+func TestDefaultAccountManager_MarkPeerDisconnected_SchedulesInactivityExpiration(t *testing.T) {
+	manager, _, err := createManager(t)
+	require.NoError(t, err, "unable to create account manager")
+
+	accountID, err := manager.GetAccountIDByUserID(context.Background(), auth.UserAuth{UserId: userID})
+	require.NoError(t, err, "unable to create an account")
+
+	key, err := wgtypes.GenerateKey()
+	require.NoError(t, err, "unable to generate WireGuard key")
+	peerPubKey := key.PublicKey().String()
+
+	_, _, _, _, err = manager.AddPeer(context.Background(), "", "", userID, &nbpeer.Peer{
+		Key:                         peerPubKey,
+		Meta:                        nbpeer.PeerSystemMeta{Hostname: "test-peer"},
+		InactivityExpirationEnabled: true,
+	}, false)
+	require.NoError(t, err, "unable to add peer")
+
+	_, err = manager.UpdateAccountSettings(context.Background(), accountID, userID, &types.Settings{
+		PeerLoginExpiration:             time.Hour,
+		PeerLoginExpirationEnabled:      true,
+		PeerInactivityExpiration:        time.Hour,
+		PeerInactivityExpirationEnabled: true,
+		Extra:                           &types.ExtraSettings{},
+	})
+	require.NoError(t, err, "expecting to update account settings successfully but got error")
+
+	// Establish a session so the matching-token disconnect is actually applied.
+	streamStartTime := time.Now().UTC()
+	err = manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, streamStartTime.UnixNano(), nil)
+	require.NoError(t, err, "unable to mark peer connected")
+
+	// Install the mock only now, so the assertion observes the disconnect, not
+	// the earlier connect.
+	scheduled := make(chan struct{}, 1)
+	manager.peerInactivityExpiry = &MockScheduler{
+		CancelFunc: func(ctx context.Context, IDs []string) {},
+		ScheduleFunc: func(ctx context.Context, in time.Duration, ID string, job func() (nextRunIn time.Duration, reschedule bool)) {
+			select {
+			case scheduled <- struct{}{}:
+			default:
+			}
+		},
+	}
+
+	err = manager.MarkPeerDisconnected(context.Background(), peerPubKey, accountID, streamStartTime.UnixNano())
+	require.NoError(t, err, "unable to mark peer disconnected")
+
+	select {
+	case <-scheduled:
+		// expected: disconnect re-armed the inactivity expiry timer
+	case <-time.After(time.Second):
+		t.Fatal("expected inactivity expiration to be rescheduled when an eligible peer disconnects")
+	}
+}
+
+func TestDefaultAccountManager_MarkPeerDisconnected_SkipsInactivityExpirationWhenDisabled(t *testing.T) {
+	manager, _, err := createManager(t)
+	require.NoError(t, err, "unable to create account manager")
+
+	accountID, err := manager.GetAccountIDByUserID(context.Background(), auth.UserAuth{UserId: userID})
+	require.NoError(t, err, "unable to create an account")
+
+	key, err := wgtypes.GenerateKey()
+	require.NoError(t, err, "unable to generate WireGuard key")
+	peerPubKey := key.PublicKey().String()
+
+	_, _, _, _, err = manager.AddPeer(context.Background(), "", "", userID, &nbpeer.Peer{
+		Key:                         peerPubKey,
+		Meta:                        nbpeer.PeerSystemMeta{Hostname: "test-peer"},
+		InactivityExpirationEnabled: true,
+	}, false)
+	require.NoError(t, err, "unable to add peer")
+
+	// Peer is eligible (SSO + inactivity enabled) but the account-level setting
+	// stays disabled, so disconnect must not schedule anything.
+	_, err = manager.UpdateAccountSettings(context.Background(), accountID, userID, &types.Settings{
+		PeerLoginExpiration:             time.Hour,
+		PeerLoginExpirationEnabled:      true,
+		PeerInactivityExpiration:        time.Hour,
+		PeerInactivityExpirationEnabled: false,
+		Extra:                           &types.ExtraSettings{},
+	})
+	require.NoError(t, err, "expecting to update account settings successfully but got error")
+
+	streamStartTime := time.Now().UTC()
+	err = manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, streamStartTime.UnixNano(), nil)
+	require.NoError(t, err, "unable to mark peer connected")
+
+	scheduled := make(chan struct{}, 1)
+	manager.peerInactivityExpiry = &MockScheduler{
+		CancelFunc: func(ctx context.Context, IDs []string) {},
+		ScheduleFunc: func(ctx context.Context, in time.Duration, ID string, job func() (nextRunIn time.Duration, reschedule bool)) {
+			select {
+			case scheduled <- struct{}{}:
+			default:
+			}
+		},
+	}
+
+	err = manager.MarkPeerDisconnected(context.Background(), peerPubKey, accountID, streamStartTime.UnixNano())
+	require.NoError(t, err, "unable to mark peer disconnected")
+
+	select {
+	case <-scheduled:
+		t.Fatal("inactivity expiration must not be scheduled while the account-level setting is disabled")
+	case <-time.After(200 * time.Millisecond):
+		// expected: nothing scheduled
+	}
+}
+
 func TestDefaultAccountManager_OnPeerDisconnected_LastSeenCheck(t *testing.T) {
 	manager, _, err := createManager(t)
 	require.NoError(t, err, "unable to create account manager")
@@ -1935,7 +2046,7 @@ func TestDefaultAccountManager_OnPeerDisconnected_LastSeenCheck(t *testing.T) {
 
 	t.Run("disconnect peer when session token matches", func(t *testing.T) {
 		streamStartTime := time.Now().UTC()
-		err = manager.MarkPeerConnected(context.Background(), peerPubKey, nil, accountID, streamStartTime.UnixNano(), nil)
+		err = manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, streamStartTime.UnixNano(), nil)
 		require.NoError(t, err, "unable to mark peer connected")
 
 		peer, err := manager.Store.GetPeerByPeerPubKey(context.Background(), store.LockingStrengthNone, peerPubKey)
@@ -1956,7 +2067,7 @@ func TestDefaultAccountManager_OnPeerDisconnected_LastSeenCheck(t *testing.T) {
 	t.Run("skip disconnect when stored session is newer (zombie stream protection)", func(t *testing.T) {
 		// Newer stream wins on connect (sets SessionStartedAt = now ns).
 		streamStartTime := time.Now().UTC()
-		err = manager.MarkPeerConnected(context.Background(), peerPubKey, nil, accountID, streamStartTime.UnixNano(), nil)
+		err = manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, streamStartTime.UnixNano(), nil)
 		require.NoError(t, err, "unable to mark peer connected")
 
 		peer, err := manager.Store.GetPeerByPeerPubKey(context.Background(), store.LockingStrengthNone, peerPubKey)
@@ -1980,7 +2091,7 @@ func TestDefaultAccountManager_OnPeerDisconnected_LastSeenCheck(t *testing.T) {
 
 	t.Run("skip stale connect when stored session is newer (blocked goroutine protection)", func(t *testing.T) {
 		node2SyncTime := time.Now().UTC()
-		err = manager.MarkPeerConnected(context.Background(), peerPubKey, nil, accountID, node2SyncTime.UnixNano(), nil)
+		err = manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, node2SyncTime.UnixNano(), nil)
 		require.NoError(t, err, "node 2 should connect peer")
 
 		peer, err := manager.Store.GetPeerByPeerPubKey(context.Background(), store.LockingStrengthNone, peerPubKey)
@@ -1990,7 +2101,7 @@ func TestDefaultAccountManager_OnPeerDisconnected_LastSeenCheck(t *testing.T) {
 			"SessionStartedAt should equal node2SyncTime token")
 
 		node1StaleSyncTime := node2SyncTime.Add(-1 * time.Minute)
-		err = manager.MarkPeerConnected(context.Background(), peerPubKey, nil, accountID, node1StaleSyncTime.UnixNano(), nil)
+		err = manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, node1StaleSyncTime.UnixNano(), nil)
 		require.NoError(t, err, "stale connect should not return error")
 
 		peer, err = manager.Store.GetPeerByPeerPubKey(context.Background(), store.LockingStrengthNone, peerPubKey)
@@ -2052,7 +2163,7 @@ func TestDefaultAccountManager_MarkPeerConnected_ConcurrentRace(t *testing.T) {
 			defer done.Done()
 			ready.Done()
 			start.Wait()
-			errs <- manager.MarkPeerConnected(context.Background(), peerPubKey, nil, accountID, token, nil)
+			errs <- manager.MarkPeerConnected(context.Background(), peerPubKey, accountID, token, nil)
 		}()
 	}
 
@@ -2093,7 +2204,7 @@ func TestDefaultAccountManager_UpdateAccountSettings_PeerLoginExpiration(t *test
 	account, err := manager.Store.GetAccount(context.Background(), accountID)
 	require.NoError(t, err, "unable to get the account")
 
-	err = manager.MarkPeerConnected(context.Background(), key.PublicKey().String(), nil, accountID, time.Now().UTC().UnixNano(), nil)
+	err = manager.MarkPeerConnected(context.Background(), key.PublicKey().String(), accountID, time.Now().UTC().UnixNano(), nil)
 	require.NoError(t, err, "unable to mark peer connected")
 
 	wg := &sync.WaitGroup{}

management/server/affected_peers_coverage_test.go

@@ -41,7 +41,7 @@ func TestAffectedPeers_DependencyCoverageMatrix(t *testing.T) {
 				_, err := s.manager.SavePolicy(ctx, s.accountID, userID, peerToResourcePolicyByGroup(s.sourceGroupID, s.resourceGroupID), true)
 				require.NoError(t, err)
 				return affectedpeers.Change{ChangedPeerIDs: []string{s.routerPeerID}},
-					[]string{s.sourcePeerID}, []string{s.unrelatedPeerID}
+					[]string{s.sourcePeerID, s.routerPeerID}, []string{s.unrelatedPeerID}
 			},
 		},
 		{
@@ -106,12 +106,8 @@ func TestAffectedPeers_DependencyCoverageMatrix(t *testing.T) {
 			change, mustContain, mustExclude := r.build(t, s, ctx)
 			affected := resolveAffected(t, s.manager.Store, s.accountID, change)
 
-			for _, id := range mustContain {
-				assert.Contains(t, affected, id, "expected peer to be affected")
-			}
-			for _, id := range mustExclude {
-				assert.NotContains(t, affected, id, "peer must not be affected")
-			}
+			assert.ElementsMatch(t, affected, mustContain, "expected peer to be affected")
+			assert.NotContains(t, affected, mustExclude, "peer must not be affected")
 		})
 	}
 }

management/server/affected_peers_router_paths_test.go

@@ -251,7 +251,9 @@ func TestAffectedPeers_E2E_UpdateResource_DestinationResourcePolicy_RefreshesSou
 	}
 }
 
-func TestAffectedPeers_E2E_UpdateResource_DisabledSiblingRouter_StillBridged(t *testing.T) {
+// A disabled sibling router routes to nobody, so updating a resource on its network
+// must NOT refresh its peer (the enabled router carries the bridge instead).
+func TestAffectedPeers_E2E_UpdateResource_DisabledSiblingRouterNotBridged(t *testing.T) {
 	s := setupRouterScenario(t, true)
 	ctx := context.Background()
 
@@ -274,13 +276,18 @@ func TestAffectedPeers_E2E_UpdateResource_DisabledSiblingRouter_StillBridged(t *
 	require.NoError(t, err)
 
 	disabledCh := s.updateManager.CreateChannel(ctx, disabledRouterPeer.ID)
-	t.Cleanup(func() { s.updateManager.CloseChannel(ctx, disabledRouterPeer.ID) })
+	enabledCh := s.updateManager.CreateChannel(ctx, s.routerPeerID)
+	t.Cleanup(func() {
+		s.updateManager.CloseChannel(ctx, disabledRouterPeer.ID)
+		s.updateManager.CloseChannel(ctx, s.routerPeerID)
+	})
 
-	settleAffectedUpdates(disabledCh)
+	settleAffectedUpdates(disabledCh, enabledCh)
 
 	done := make(chan struct{})
 	go func() {
-		peerShouldReceiveUpdate(t, disabledCh)
+		peerShouldReceiveUpdate(t, enabledCh)
+		peerShouldNotReceiveUpdate(t, disabledCh)
 		close(done)
 	}()
 
@@ -298,7 +305,7 @@ func TestAffectedPeers_E2E_UpdateResource_DisabledSiblingRouter_StillBridged(t *
 	select {
 	case <-done:
 	case <-time.After(peerUpdateTimeout):
-		t.Error("timeout: resource update did not refresh the disabled sibling router's peer")
+		t.Error("timeout")
 	}
 }
 

management/server/affected_peers_router_test.go

@@ -682,6 +682,9 @@ func TestAffectedPeers_AllRoutingPeers_Network(t *testing.T) {
 	assert.Contains(t, affected, secondRouterPeer.ID, "second routing peer on the same network must also be affected")
 }
 
+// A disabled router in the snapshot routes to nobody, so it is skipped when the
+// walk scans existing account data: a policy edit still folds the literal source
+// group, but not the disabled router's peer.
 func TestAffectedPeers_DisabledRouter(t *testing.T) {
 	s := setupRouterScenario(t, true)
 	ctx := context.Background()
@@ -694,11 +697,13 @@ func TestAffectedPeers_DisabledRouter(t *testing.T) {
 
 	affected := s.resolvePolicyAffected(ctx, peerToResourcePolicyByGroup(s.sourceGroupID, s.resourceGroupID))
 
-	assert.Contains(t, affected, s.sourcePeerID, "source peer must be affected")
-	assert.Contains(t, affected, s.routerPeerID,
-		"disabled router's peer must still be affected: Enabled must not gate affected-peers")
+	assert.Contains(t, affected, s.sourcePeerID, "source peer (literal policy source group) must be affected")
+	assert.NotContains(t, affected, s.routerPeerID,
+		"a disabled router routes to nobody, so its peer must not be folded from snapshot data")
 }
 
+// A disabled resource in the snapshot is skipped: the policy edit still folds the
+// literal source group, but the resource no longer bridges to its network's router.
 func TestAffectedPeers_DisabledResource(t *testing.T) {
 	s := setupRouterScenario(t, true)
 	ctx := context.Background()
@@ -710,9 +715,9 @@ func TestAffectedPeers_DisabledResource(t *testing.T) {
 
 	affected := s.resolvePolicyAffected(ctx, peerToResourcePolicyByGroup(s.sourceGroupID, s.resourceGroupID))
 
-	assert.Contains(t, affected, s.sourcePeerID, "source peer must be affected")
-	assert.Contains(t, affected, s.routerPeerID,
-		"disabled resource must still resolve the routing peer: Enabled must not gate affected-peers")
+	assert.Contains(t, affected, s.sourcePeerID, "source peer (literal policy source group) must be affected")
+	assert.NotContains(t, affected, s.routerPeerID,
+		"a disabled resource routes to nobody, so its network's router must not be folded from snapshot data")
 }
 
 func TestAffectedPeers_DisabledRule(t *testing.T) {

management/server/affected_peers_test.go

@@ -96,33 +96,54 @@ func affectedGroupID(i int) string   { return fmt.Sprintf("affected-grp-%d", i)
 func affectedGroupName(i int) string { return fmt.Sprintf("AffectedGroup%d", i) }
 
 func TestCollectGroupChange_PolicyLinked(t *testing.T) {
-	manager, s, accountID, _, groupIDs := setupAffectedPeersTest(t)
+	manager, s, accountID, peerIDs, groupIDs := setupAffectedPeersTest(t)
 	ctx := context.Background()
 
 	_, err := manager.SavePolicy(ctx, accountID, userID, &types.Policy{
 		Enabled: true,
 		Rules: []*types.PolicyRule{
 			{
-				Enabled:       true,
-				Sources:       []string{groupIDs[0]},
-				Destinations:  []string{groupIDs[1]},
-				Bidirectional: true,
-				Action:        types.PolicyTrafficActionAccept,
+				Enabled:             true,
+				Sources:             []string{groupIDs[0]},
+				Destinations:        []string{groupIDs[1]},
+				SourceResource:      types.Resource{ID: peerIDs[0], Type: types.ResourceTypePeer},
+				DestinationResource: types.Resource{ID: peerIDs[1], Type: types.ResourceTypePeer},
+				Bidirectional:       true,
+				Action:              types.PolicyTrafficActionAccept,
+			},
+			{
+				Enabled:             true,
+				Sources:             []string{groupIDs[0]},
+				Destinations:        []string{groupIDs[1]},
+				SourceResource:      types.Resource{ID: peerIDs[2], Type: types.ResourceTypeHost},
+				DestinationResource: types.Resource{ID: peerIDs[3], Type: types.ResourceTypeHost},
+				Bidirectional:       true,
+				Action:              types.PolicyTrafficActionAccept,
+			},
+			{
+				Enabled:             true,
+				Sources:             []string{groupIDs[0]},
+				Destinations:        []string{groupIDs[1]},
+				SourceResource:      types.Resource{ID: "", Type: types.ResourceTypePeer},
+				DestinationResource: types.Resource{ID: "", Type: types.ResourceTypePeer},
+				Bidirectional:       true,
+				Action:              types.PolicyTrafficActionAccept,
 			},
 		},
 	}, true)
 	require.NoError(t, err)
 
-	groups, _ := collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[0]})
-	assert.Contains(t, groups, groupIDs[0])
-	assert.Contains(t, groups, groupIDs[1])
+	groups, directPeers := collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[0]})
+	assert.ElementsMatch(t, groups, []string{groupIDs[0], groupIDs[1]})
+	assert.ElementsMatch(t, directPeers, []string{peerIDs[1]})
 
-	groups, _ = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[1]})
-	assert.Contains(t, groups, groupIDs[0])
-	assert.Contains(t, groups, groupIDs[1])
+	groups, directPeers = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[1]})
+	assert.ElementsMatch(t, groups, []string{groupIDs[0], groupIDs[1]})
+	assert.ElementsMatch(t, directPeers, []string{peerIDs[0]})
 
-	groups, _ = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[2]})
+	groups, directPeers = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[2]})
 	assert.Empty(t, groups)
+	assert.Empty(t, directPeers)
 }
 
 func TestCollectGroupChange_PolicyWithDirectPeerResource(t *testing.T) {
@@ -133,20 +154,44 @@ func TestCollectGroupChange_PolicyWithDirectPeerResource(t *testing.T) {
 		Enabled: true,
 		Rules: []*types.PolicyRule{
 			{
-				Enabled:        true,
-				Sources:        []string{groupIDs[0]},
-				SourceResource: types.Resource{ID: peerIDs[3], Type: types.ResourceTypePeer},
-				Destinations:   []string{groupIDs[1]},
-				Action:         types.PolicyTrafficActionAccept,
+				Enabled:             true,
+				Sources:             []string{groupIDs[0]},
+				SourceResource:      types.Resource{ID: peerIDs[3], Type: types.ResourceTypePeer},
+				DestinationResource: types.Resource{ID: peerIDs[4], Type: types.ResourceTypePeer},
+				Destinations:        []string{groupIDs[1]},
+				Action:              types.PolicyTrafficActionAccept,
+			},
+			{
+				Enabled:             true,
+				Sources:             []string{groupIDs[0]},
+				SourceResource:      types.Resource{ID: peerIDs[1], Type: types.ResourceTypeHost},
+				DestinationResource: types.Resource{ID: peerIDs[2], Type: types.ResourceTypeHost},
+				Destinations:        []string{groupIDs[1]},
+				Action:              types.PolicyTrafficActionAccept,
+			},
+			{
+				Enabled:             true,
+				Sources:             []string{groupIDs[0]},
+				SourceResource:      types.Resource{ID: "", Type: types.ResourceTypePeer},
+				DestinationResource: types.Resource{ID: "", Type: types.ResourceTypePeer},
+				Destinations:        []string{groupIDs[1]},
+				Action:              types.PolicyTrafficActionAccept,
 			},
 		},
 	}, true)
 	require.NoError(t, err)
 
 	groups, directPeers := collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[0]})
-	assert.Contains(t, groups, groupIDs[0])
-	assert.Contains(t, groups, groupIDs[1])
-	assert.Contains(t, directPeers, peerIDs[3])
+	assert.ElementsMatch(t, groups, []string{groupIDs[0], groupIDs[1]})
+	assert.ElementsMatch(t, directPeers, []string{peerIDs[4]})
+
+	groups, directPeers = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[1]})
+	assert.ElementsMatch(t, groups, []string{groupIDs[0], groupIDs[1]})
+	assert.ElementsMatch(t, directPeers, []string{peerIDs[3]})
+
+	groups, directPeers = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[2]})
+	assert.Empty(t, groups)
+	assert.Empty(t, directPeers)
 }
 
 func TestCollectGroupChange_PolicyWithNonPeerResource_NoDirectPeers(t *testing.T) {
@@ -168,8 +213,7 @@ func TestCollectGroupChange_PolicyWithNonPeerResource_NoDirectPeers(t *testing.T
 	require.NoError(t, err)
 
 	groups, directPeers := collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[0]})
-	assert.Contains(t, groups, groupIDs[0])
-	assert.Contains(t, groups, groupIDs[1])
+	assert.ElementsMatch(t, groups, []string{groupIDs[0], groupIDs[1]})
 	assert.Empty(t, directPeers, "non-peer resources should not produce direct peer IDs")
 }
 
@@ -294,6 +338,7 @@ func TestCollectGroupChange_NetworkRouterLinked(t *testing.T) {
 		AccountID:  accountID,
 		PeerGroups: []string{groupIDs[0]},
 		Peer:       peerIDs[3],
+		Enabled:    true,
 	})
 	require.NoError(t, err)
 
@@ -324,6 +369,7 @@ func TestCollectGroupChange_NetworkRouterPeerOnlyNoGroups(t *testing.T) {
 		NetworkID: net1.ID,
 		AccountID: accountID,
 		Peer:      peerIDs[4],
+		Enabled:   true,
 	})
 	require.NoError(t, err)
 
@@ -373,17 +419,11 @@ func TestCollectGroupChange_MultipleEntities(t *testing.T) {
 	require.NoError(t, err)
 
 	groups, directPeers := collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[0]})
-	assert.Contains(t, groups, groupIDs[0])
-	assert.Contains(t, groups, groupIDs[1])
-	assert.NotContains(t, groups, groupIDs[2])
-	assert.NotContains(t, groups, groupIDs[3])
+	assert.ElementsMatch(t, groups, []string{groupIDs[0], groupIDs[1]})
 	assert.Empty(t, directPeers)
 
 	groups, directPeers = collectGroupChangeAffectedGroups(ctx, s, accountID, []string{groupIDs[3]})
-	assert.Contains(t, groups, groupIDs[2])
-	assert.Contains(t, groups, groupIDs[3])
-	assert.NotContains(t, groups, groupIDs[0])
-	assert.NotContains(t, groups, groupIDs[1])
+	assert.ElementsMatch(t, groups, []string{groupIDs[2], groupIDs[3]})
 	assert.Empty(t, directPeers)
 }
 
@@ -452,8 +492,9 @@ func TestResolveAffectedPeers_PolicyBetweenTwoGroups(t *testing.T) {
 	result = manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[1]})
 	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[1]}, result)
 
+	// peerIDs[2] is unrelated to the route; only its own map can change.
 	result = manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[2]})
-	assert.Empty(t, result)
+	assert.ElementsMatch(t, []string{peerIDs[2]}, result)
 }
 
 func TestResolveAffectedPeers_PolicyThreeGroups(t *testing.T) {
@@ -474,7 +515,7 @@ func TestResolveAffectedPeers_PolicyThreeGroups(t *testing.T) {
 	require.NoError(t, err)
 
 	result := manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[0]})
-	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[1], peerIDs[2]}, result)
+	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[2]}, result)
 }
 
 func TestResolveAffectedPeers_RoutePeerGroups(t *testing.T) {
@@ -506,8 +547,9 @@ func TestResolveAffectedPeers_RoutePeerGroups(t *testing.T) {
 	result = manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[1]})
 	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[1]}, result)
 
+	// peerIDs[2] is in no policy; only its own map can change, so it refreshes itself.
 	result = manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[2]})
-	assert.Empty(t, result)
+	assert.ElementsMatch(t, []string{peerIDs[2]}, result)
 }
 
 func TestResolveAffectedPeers_RouteWithDirectPeer(t *testing.T) {
@@ -564,9 +606,9 @@ func TestResolveAffectedPeers_RouteWithAccessControlGroups(t *testing.T) {
 	result := manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[2]})
 	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[1], peerIDs[2]}, result)
 
-	// peer3 is unrelated
+	// peer3 is unrelated to the route; only its own map can change.
 	result = manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[3]})
-	assert.Empty(t, result)
+	assert.ElementsMatch(t, []string{peerIDs[3]}, result)
 }
 
 func TestResolveAffectedPeers_NetworkRouter(t *testing.T) {
@@ -587,6 +629,7 @@ func TestResolveAffectedPeers_NetworkRouter(t *testing.T) {
 		AccountID:  accountID,
 		PeerGroups: []string{groupIDs[0]},
 		Peer:       peerIDs[3],
+		Enabled:    true,
 	})
 	require.NoError(t, err)
 
@@ -659,9 +702,13 @@ func TestResolveAffectedPeers_PeerInMultipleGroups(t *testing.T) {
 	}, true)
 	require.NoError(t, err)
 
-	// peer0 is in group0 AND group1, so both policies apply
+	// peer0 is in group0 AND group1, so both policies apply. A peer change folds
+	// only the changed peer plus the opposite side of each rule: group2 (peer2) via
+	// the group0 policy and group3 (peer3) via the group1 policy. peer1, a co-member
+	// of group1, is a sibling of the changed peer and must NOT refresh.
 	result := manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[0]})
-	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[1], peerIDs[2], peerIDs[3]}, result)
+	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[2], peerIDs[3]}, result)
+	assert.NotContains(t, result, peerIDs[1], "co-member of the changed peer's group must not refresh")
 }
 
 func TestResolveAffectedPeers_MultipleChangedPeers(t *testing.T) {
@@ -697,7 +744,7 @@ func TestResolveAffectedPeers_MultipleChangedPeers(t *testing.T) {
 	require.NoError(t, err)
 
 	result := manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[0], peerIDs[2]})
-	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[1], peerIDs[2], peerIDs[3]}, result)
+	assert.ElementsMatch(t, []string{peerIDs[0], peerIDs[2], peerIDs[1], peerIDs[3]}, result)
 }
 
 func TestResolveAffectedPeers_SharedGroupAcrossPolicyAndRoute(t *testing.T) {
@@ -854,8 +901,9 @@ func TestAffectedPeers_IsolatedPolicies(t *testing.T) {
 	assert.NotContains(t, result, peerIDs[0])
 	assert.NotContains(t, result, peerIDs[1])
 
+	// peerIDs[4] is in neither isolated policy; only its own map can change.
 	result = manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[4]})
-	assert.Empty(t, result)
+	assert.ElementsMatch(t, []string{peerIDs[4]}, result)
 }
 
 func TestAffectedPeers_IsolatedRouteAndPolicy(t *testing.T) {
@@ -977,12 +1025,13 @@ func TestAffectedPeers_GroupUpdateOnlyAffectsLinkedPeers(t *testing.T) {
 	})
 }
 
-func TestAffectedPeers_UnlinkedGroupChange_NoUpdates(t *testing.T) {
+// A peer in no policy/route refreshes only itself — no other peer is affected.
+func TestAffectedPeers_UnlinkedPeerChange_RefreshesSelfOnly(t *testing.T) {
 	manager, s, accountID, peerIDs, _ := setupAffectedPeersTest(t)
 	ctx := context.Background()
 
 	result := manager.resolveAffectedPeersForPeerChanges(ctx, s, accountID, []string{peerIDs[0]})
-	assert.Empty(t, result)
+	assert.ElementsMatch(t, []string{peerIDs[0]}, result)
 }
 
 // TestAffectedPeers_PolicyChange_UnrelatedPeerNoUpdate verifies that creating/deleting a
@@ -1332,6 +1381,7 @@ func TestAffectedPeers_NetworkRouterUnlinkedPeerNoUpdate(t *testing.T) {
 		NetworkID:  net1.ID,
 		AccountID:  accountID,
 		PeerGroups: []string{"nr-grpA"},
+		Enabled:    true,
 	})
 	require.NoError(t, err)
 
@@ -1755,7 +1805,9 @@ func TestCollectAffectedFromProxyServices_GroupContainingTargetPeerChanged(t *te
 	assert.Contains(t, directPeers, peerIDs[1], "target peer must be refreshed")
 }
 
-func TestCollectAffectedFromProxyServices_DisabledServiceStillMatches(t *testing.T) {
+// A disabled service in the snapshot proxies nothing, so it is skipped: a changed
+// target peer does not pull in the service's proxy peer.
+func TestCollectAffectedFromProxyServices_DisabledServiceSkipped(t *testing.T) {
 	manager, s, accountID, peerIDs, _ := setupAffectedPeersTest(t)
 	ctx := context.Background()
 
@@ -1781,8 +1833,7 @@ func TestCollectAffectedFromProxyServices_DisabledServiceStillMatches(t *testing
 	require.NoError(t, s.CreateService(ctx, svc))
 
 	_, directPeers := collectPeerChangeAffectedGroups(ctx, manager.Store, accountID, nil, []string{peerIDs[1]})
-	assert.Contains(t, directPeers, peerIDs[0], "disabled service should still trigger a refresh so peers are ready when re-enabled")
-	assert.Contains(t, directPeers, peerIDs[1], "disabled target should still trigger a refresh")
+	assert.NotContains(t, directPeers, peerIDs[0], "a disabled service proxies nothing, so its proxy peer must not be folded")
 }
 
 func TestCollectAffectedFromProxyServices_NonPeerTargetType(t *testing.T) {

management/server/affectedpeers/resolver.go

@@ -6,7 +6,12 @@
 //     and before a delete/removal severs the old state).
 //   - Snapshot.Expand: in-memory walk, no store access. Run AFTER the tx commits.
 //
-// Enabled is never consulted: toggling it is itself an observable change.
+// Enabled handling differs by source. Disabled objects in the SNAPSHOT (existing
+// account policies/resources/routers/routes/proxy services and their rules/targets)
+// route to nobody and are skipped — they cannot affect any peer's map. Objects in
+// the CHANGE itself are processed regardless of Enabled, so disabling one still
+// refreshes the peers that lose access (the toggle is the observable change, and the
+// update carries the old∪new state).
 package affectedpeers
 
 import (
@@ -61,7 +66,8 @@ func Load(ctx context.Context, s store.Store, accountID string, c Change) (*Snap
 // loadCollections reads the policy/route/nameserver/dns/router/resource/proxy
 // collections a Change can touch, gated to what the walk needs.
 func (snap *Snapshot) loadCollections(ctx context.Context, s store.Store, accountID string, c Change) error {
-	hasGroupOrPeerChange := len(c.ChangedGroupIDs) > 0 || len(c.ChangedPeerIDs) > 0 || len(c.Resources) > 0
+	// LinkGroups drive the same policy/route/dns walk as a changed group or peer.
+	hasGroupOrPeerChange := len(c.ChangedGroupIDs) > 0 || len(c.ChangedPeerIDs) > 0 || len(c.LinkGroups) > 0 || len(c.Resources) > 0
 	hasNetworkObject := len(c.Routers) > 0 || len(c.Resources) > 0 || len(c.Networks) > 0
 	// the resource<->router bridge can fire for any of these
 	needsRoutersResources := hasGroupOrPeerChange || len(c.PostureCheckIDs) > 0 || len(c.Policies) > 0 || hasNetworkObject
@@ -76,7 +82,7 @@ func (snap *Snapshot) loadCollections(ctx context.Context, s store.Store, accoun
 			return err
 		}
 	}
-	if len(c.ChangedGroupIDs) > 0 || len(c.ChangedPeerIDs) > 0 {
+	if len(c.ChangedGroupIDs) > 0 || len(c.ChangedPeerIDs) > 0 || len(c.LinkGroups) > 0 {
 		if err := snap.loadDNS(ctx, s, accountID); err != nil {
 			return err
 		}
@@ -174,6 +180,24 @@ type Change struct {
 	// folded in — but only when the group is linked (an unlinked group has no map
 	// impact), matching how current members are handled.
 	RemovedPeersByGroup map[string][]string
+
+	// OutputPeerIDs are peers folded straight into the result without seeding their
+	// group memberships into the walk. Use for the peer whose group membership changed:
+	// the peer itself must refresh, but its OTHER groups did not change, so they must
+	// not be walked. Contrast ChangedPeerIDs, which seeds ALL of the peer's groups
+	// (correct when the peer's own attributes changed, e.g. IP/status).
+	OutputPeerIDs []string
+
+	// LinkGroups are groups used ONLY to match policies/routes/routers and walk to the
+	// OPPOSITE side — they are never expanded to their own members. Use this when a
+	// peer's group membership changed: pass the peer in ChangedPeerIDs and its
+	// group(s) here. The opposite side of the policies the group participates in
+	// refreshes, but the group's other members (siblings) do not — nothing changed for
+	// them. For an intra-group policy (A→A) the opposite side IS the group, so its
+	// members still refresh via the opposite-side fold, exactly when they genuinely
+	// gain/lose the changed peer. Unlike ChangedGroupIDs, a LinkGroup is not added to
+	// the output, so a one-sided membership change never wakes the whole group.
+	LinkGroups []string
 }
 
 func (c Change) isEmpty() bool {
@@ -186,7 +210,9 @@ func (c Change) isEmpty() bool {
 		len(c.Networks) == 0 &&
 		len(c.PostureCheckIDs) == 0 &&
 		len(c.DistributionGroupIDs) == 0 &&
-		len(c.RemovedPeersByGroup) == 0
+		len(c.RemovedPeersByGroup) == 0 &&
+		len(c.LinkGroups) == 0 &&
+		len(c.OutputPeerIDs) == 0
 }
 
 // Expand returns the deduplicated affected peer IDs from the preloaded Snapshot,
@@ -197,8 +223,8 @@ func (snap *Snapshot) Expand(ctx context.Context, accountID string, c Change) []
 		return nil
 	}
 	r := newResolver(ctx, snap, accountID, c)
-	log.WithContext(ctx).Tracef("affectedpeers expand start: account=%s changedGroups=%v changedPeers=%v policies=%d routes=%d routers=%d resources=%d networks=%d postureChecks=%v distributionGroups=%v",
-		accountID, c.ChangedGroupIDs, c.ChangedPeerIDs, len(c.Policies), len(c.Routes), len(c.Routers), len(c.Resources), len(c.Networks), c.PostureCheckIDs, c.DistributionGroupIDs)
+	log.WithContext(ctx).Tracef("affectedpeers expand start: account=%s changedGroups=%v changedPeers=%v linkGroups=%v policies=%d routes=%d routers=%d resources=%d networks=%d postureChecks=%v distributionGroups=%v",
+		accountID, c.ChangedGroupIDs, c.ChangedPeerIDs, c.LinkGroups, len(c.Policies), len(c.Routes), len(c.Routers), len(c.Resources), len(c.Networks), c.PostureCheckIDs, c.DistributionGroupIDs)
 	r.walk()
 	return r.expand()
 }
@@ -216,57 +242,84 @@ func Collect(ctx context.Context, s store.Store, accountID string, c Change) (gr
 	}
 	r := newResolver(ctx, snap, accountID, c)
 	r.walk()
-	return setToSlice(r.groupSet), setToSlice(r.peerSet)
+	return setToSlice(r.affectedGroups), setToSlice(r.affectedPeers)
 }
 
 func newResolver(ctx context.Context, snap *Snapshot, accountID string, c Change) *resolver {
 	r := &resolver{
-		ctx:             ctx,
-		snap:            snap,
-		accountID:       accountID,
-		change:          c,
-		changedGroupSet: toSet(c.ChangedGroupIDs),
-		changedPeerSet:  toSet(c.ChangedPeerIDs),
-		groupSet:        make(map[string]struct{}),
-		peerSet:         make(map[string]struct{}),
-		networkIDs:      make(map[string]struct{}),
+		ctx:            ctx,
+		snap:           snap,
+		accountID:      accountID,
+		change:         c,
+		linkGroups:     toSet(c.ChangedGroupIDs),
+		outputGroups:   toSet(c.ChangedGroupIDs),
+		changedPeers:   toSet(c.ChangedPeerIDs),
+		affectedGroups: make(map[string]struct{}),
+		affectedPeers:  make(map[string]struct{}),
 	}
+	// LinkGroups match policies/routes to find the opposite side but are NOT output:
+	// they go into linkGroups only, never outputGroups, so their members never fold in.
+	addAll(r.linkGroups, c.LinkGroups)
 	// Resolve each changed peer to its groups here so callers pass only ChangedPeerIDs.
 	r.seedChangedGroupsFromPeers()
-	r.matchedPolicies = append(r.matchedPolicies, c.Policies...)
 	return r
 }
 
-// seedChangedGroupsFromPeers adds each changed peer's groups to changedGroupSet so
+// seedChangedGroupsFromPeers adds each changed peer's groups to linkGroups so
 // the group-driven walkers fire for memberships, not just direct peer references.
+// These seeded groups are for MATCHING only — folding the changed entity's own
+// side is gated on outputGroups (the caller-reported groups), so a seeded group
+// never folds its whole membership; only the changed peer itself folds in.
 func (r *resolver) seedChangedGroupsFromPeers() {
-	if len(r.changedPeerSet) == 0 {
+	if len(r.changedPeers) == 0 {
 		return
 	}
 	for groupID, members := range r.snap.groupPeers {
-		for pID := range r.changedPeerSet {
+		for pID := range r.changedPeers {
 			if _, ok := members[pID]; ok {
-				r.changedGroupSet[groupID] = struct{}{}
+				r.linkGroups[groupID] = struct{}{}
 				break
 			}
 		}
 	}
 }
 
+// policySide selects which side of a policy rule to walk.
+type policySide int
+
+const (
+	sideSource policySide = iota
+	sideDestination
+)
+
+func (s policySide) opposite() policySide {
+	if s == sideSource {
+		return sideDestination
+	}
+	return sideSource
+}
+
+// walk resolves affected peers in two buckets, by how far each change propagates.
+//
+// BOTH-SIDES — the rule itself changed (an explicit policy edit, or a policy whose
+// posture check changed). Source AND destination refresh, so each such policy is
+// walked on both sides.
+//
+// OPPOSITE-SIDE — an endpoint moved but no rule changed. For each policy the change
+// touches we fold only the side AWAY from the change:
+//   - a changed peer/group sits ON a policy side -> fold the opposite side;
+//   - a changed router/resource/network sits on a NETWORK -> fold the SOURCE side of
+//     the policies whose destination reaches it (and the routers it implies).
+//
+// Routes, nameserver groups, DNS and embedded-proxy services distribute to their own
+// member peers, outside the policy graph, and are folded here too.
 func (r *resolver) walk() {
-	r.collectFromExplicitPolicies()
-	r.collectFromExplicitRoutes(r.change.Routes)
-	r.collectFromExplicitRouters(r.change.Routers)
-	r.collectFromExplicitResources(r.change.Resources)
-	r.collectFromExplicitNetworks(r.change.Networks)
-	r.collectFromPostureChecks(r.change.PostureCheckIDs)
+	for _, policy := range r.bothSidesPolicies() {
+		r.foldPolicySide(policy, sideSource)
+		r.foldPolicySide(policy, sideDestination)
+	}
 
-	// Distribution groups (nameserver/DNS) affect only their member peers: fold them
-	// straight into groupSet so expand() maps them to members, without the policy/
-	// route walk that changedGroupSet would trigger.
-	addAll(r.groupSet, r.change.DistributionGroupIDs)
-
-	if len(r.changedGroupSet) > 0 || len(r.changedPeerSet) > 0 {
+	if len(r.linkGroups) > 0 || len(r.changedPeers) > 0 {
 		r.collectFromPolicies()
 		r.collectFromRoutes()
 		r.collectFromNameServers()
@@ -275,7 +328,31 @@ func (r *resolver) walk() {
 		r.collectFromProxyServices()
 	}
 
-	r.collectResourceRouterBridge()
+	r.collectFromChangedRoutes(r.change.Routes)
+	r.collectFromChangedRouters(r.change.Routers)
+	r.collectFromChangedResources(r.change.Resources)
+	r.collectFromChangedNetworks(r.change.Networks)
+
+	// The explicitly changed peers always refresh their own maps. OnPeersUpdated only
+	// refreshes the resolver's output (it ignores the separately-passed changed peers),
+	// so the changed peer reaches its own new map only via here. An offline/deleted
+	// peer in the set is filtered downstream (filterConnectedAffectedPeers).
+	addAll(r.affectedPeers, setToSlice(r.changedPeers))
+	// OutputPeerIDs refresh themselves too, but unlike changedPeers their group
+	// memberships were not seeded into the walk (only the changed group was).
+	addAll(r.affectedPeers, r.change.OutputPeerIDs)
+
+	// Distribution groups (nameserver/DNS) affect only their member peers: fold them
+	// straight into affectedGroups so expand() maps them to members, without the
+	// policy/route walk that linkGroups would trigger.
+	addAll(r.affectedGroups, r.change.DistributionGroupIDs)
+}
+
+// bothSidesPolicies are the policies whose rule changed: the explicitly edited ones
+// plus those gated by a changed posture check. walk folds both their sides.
+func (r *resolver) bothSidesPolicies() []*types.Policy {
+	policies := append([]*types.Policy(nil), r.change.Policies...)
+	return r.appendPoliciesForPostureChecks(policies, r.change.PostureCheckIDs)
 }
 
 type resolver struct {
@@ -284,27 +361,71 @@ type resolver struct {
 	accountID string
 	change    Change
 
-	changedGroupSet map[string]struct{}
-	changedPeerSet  map[string]struct{}
+	// Inputs — what changed. Set once at construction, read-only during the walk
+	// (except linkGroups, which collectFromExplicitResources also seeds).
+	//
+	// linkGroups is the MATCH set: caller-changed groups ∪ the groups of changed
+	// peers ∪ changed-resource groups. A rule/route/router matches the change when
+	// one of its groups is here — used only to find the opposite side to fold.
+	//
+	// outputGroups is the FOLD-WHOLE-GROUP set: ONLY Change.ChangedGroupIDs. When a
+	// matched group is here, its whole membership is affected. A peer-seeded group
+	// is in linkGroups but NOT outputGroups, so it folds only the changed peer
+	// (changedPeers), never its siblings.
+	linkGroups   map[string]struct{}
+	outputGroups map[string]struct{}
+	changedPeers map[string]struct{}
 
-	groupSet map[string]struct{}
-	peerSet  map[string]struct{}
-
-	matchedPolicies []*types.Policy
-	networkIDs      map[string]struct{}
+	// Outputs — the answer. The only sets the walk accumulates into. affectedGroups
+	// is expanded to its member peers in expand().
+	affectedGroups map[string]struct{}
+	affectedPeers  map[string]struct{}
 }
 
-func (r *resolver) policies() []*types.Policy { return r.snap.policies }
+// policies returns the account's ENABLED policies from the snapshot. Disabled
+// policies grant no access, so the walk skips them when scanning existing account
+// data. Explicitly changed policies (Change.Policies, via bothSidesPolicies) are
+// processed regardless of Enabled, so disabling one still refreshes its peers.
+func (r *resolver) policies() []*types.Policy {
+	enabled := make([]*types.Policy, 0, len(r.snap.policies))
+	for _, policy := range r.snap.policies {
+		if policy != nil && policy.Enabled {
+			enabled = append(enabled, policy)
+		}
+	}
+	return enabled
+}
 
-func (r *resolver) networkResources() []*resourceTypes.NetworkResource { return r.snap.resources }
+// networkResources / networkRouters return the account's ENABLED resources/routers
+// from the snapshot. Disabled objects route to nobody, so the walk skips them when
+// it scans existing account data. The explicitly changed objects in the Change are
+// processed regardless of Enabled (collectFromChanged*), so disabling one still
+// refreshes the peers that lose access.
+func (r *resolver) networkResources() []*resourceTypes.NetworkResource {
+	enabled := make([]*resourceTypes.NetworkResource, 0, len(r.snap.resources))
+	for _, resource := range r.snap.resources {
+		if resource.Enabled {
+			enabled = append(enabled, resource)
+		}
+	}
+	return enabled
+}
 
-func (r *resolver) networkRouters() []*routerTypes.NetworkRouter { return r.snap.routers }
+func (r *resolver) networkRouters() []*routerTypes.NetworkRouter {
+	enabled := make([]*routerTypes.NetworkRouter, 0, len(r.snap.routers))
+	for _, router := range r.snap.routers {
+		if router.Enabled {
+			enabled = append(enabled, router)
+		}
+	}
+	return enabled
+}
 
 // peerIDsForGroups maps a group set to its member peer IDs via the preloaded index.
-func (r *resolver) peerIDsForGroups(groupSet map[string]struct{}) []string {
+func (r *resolver) peerIDsForGroups(groups map[string]struct{}) []string {
 	seen := make(map[string]struct{})
 	var ids []string
-	for gID := range groupSet {
+	for gID := range groups {
 		for pID := range r.snap.groupPeers[gID] {
 			if _, ok := seen[pID]; ok {
 				continue
@@ -317,25 +438,25 @@ func (r *resolver) peerIDsForGroups(groupSet map[string]struct{}) []string {
 }
 
 func (r *resolver) expand() []string {
-	peerIDs := r.peerIDsForGroups(r.groupSet)
+	peerIDs := r.peerIDsForGroups(r.affectedGroups)
 
 	log.WithContext(r.ctx).Tracef("affectedpeers expand: account=%s affectedGroups=%v -> %d group-member peers; direct peers=%v",
-		r.accountID, setToSlice(r.groupSet), len(peerIDs), setToSlice(r.peerSet))
+		r.accountID, setToSlice(r.affectedGroups), len(peerIDs), setToSlice(r.affectedPeers))
 
 	seen := make(map[string]struct{}, len(peerIDs))
 	for _, id := range peerIDs {
 		seen[id] = struct{}{}
 	}
-	for id := range r.peerSet {
+	for id := range r.affectedPeers {
 		if _, ok := seen[id]; !ok {
 			peerIDs = append(peerIDs, id)
 			seen[id] = struct{}{}
 		}
 	}
 
-	// Fold in removed peers only when their group is linked (in groupSet).
+	// Fold in removed peers only when their group is linked (in affectedGroups).
 	for groupID, removed := range r.change.RemovedPeersByGroup {
-		if _, linked := r.groupSet[groupID]; !linked {
+		if _, linked := r.affectedGroups[groupID]; !linked {
 			continue
 		}
 		for _, id := range removed {
@@ -351,169 +472,309 @@ func (r *resolver) expand() []string {
 	return peerIDs
 }
 
-func (r *resolver) collectFromExplicitPolicies() {
-	for _, policy := range r.matchedPolicies {
-		if policy == nil {
-			continue
-		}
-		log.WithContext(r.ctx).Tracef("collectFromExplicitPolicies: changed policy %s (%s) -> folding rule groups %v + direct peers",
-			policy.ID, policy.Name, policy.RuleGroups())
-		addAll(r.groupSet, policy.RuleGroups())
-		collectPolicyDirectPeers(policy, r.peerSet)
+// ruleSideGroups / ruleSideResource return the groups and the resource on the given
+// side of a rule.
+func ruleSideGroups(rule *types.PolicyRule, side policySide) []string {
+	if side == sideDestination {
+		return rule.Destinations
 	}
+	return rule.Sources
 }
 
-func (r *resolver) collectFromExplicitRoutes(routes []*route.Route) {
-	for _, rt := range routes {
-		if rt == nil {
-			continue
-		}
-		log.WithContext(r.ctx).Tracef("collectFromExplicitRoutes: changed route %s -> folding groups=%v peerGroups=%v accessControlGroups=%v peer=%q",
-			rt.ID, rt.Groups, rt.PeerGroups, rt.AccessControlGroups, rt.Peer)
-		addAll(r.groupSet, rt.Groups, rt.PeerGroups, rt.AccessControlGroups)
-		if rt.Peer != "" {
-			r.peerSet[rt.Peer] = struct{}{}
-		}
+func ruleSideResource(rule *types.PolicyRule, side policySide) types.Resource {
+	if side == sideDestination {
+		return rule.DestinationResource
 	}
+	return rule.SourceResource
 }
 
-// collectFromExplicitRouters folds changed routers' peers and marks their networks
-// for the bridge. Passing the old router keeps a repointed router's previous peers
-// affected without a post-commit read.
-func (r *resolver) collectFromExplicitRouters(routers []*routerTypes.NetworkRouter) {
-	for _, router := range routers {
-		if router == nil {
-			continue
-		}
-		log.WithContext(r.ctx).Tracef("collectFromExplicitRouters: changed router %s on network %s -> folding peerGroups=%v peer=%q and marking network for source bridge",
-			router.ID, router.NetworkID, router.PeerGroups, router.Peer)
-		addAll(r.groupSet, router.PeerGroups)
-		if router.Peer != "" {
-			r.peerSet[router.Peer] = struct{}{}
-		}
-		if router.NetworkID != "" {
-			r.networkIDs[router.NetworkID] = struct{}{}
-		}
-	}
-}
-
-// collectFromExplicitResources marks changed resources' networks for the bridge and
-// treats their group IDs as changed, so policies targeting the resource via a
-// now-detached (old) group still refresh.
-func (r *resolver) collectFromExplicitResources(resources []*resourceTypes.NetworkResource) {
-	for _, resource := range resources {
-		if resource == nil {
-			continue
-		}
-		log.WithContext(r.ctx).Tracef("collectFromExplicitResources: changed resource %s on network %s -> marking network for bridge and treating groups %v as changed",
-			resource.ID, resource.NetworkID, resource.GroupIDs)
-		addAll(r.changedGroupSet, resource.GroupIDs)
-		if resource.NetworkID != "" {
-			r.networkIDs[resource.NetworkID] = struct{}{}
-		}
-	}
-}
-
-// collectFromExplicitNetworks marks changed networks for the bridge. A network has
-// no groups/peers of its own.
-func (r *resolver) collectFromExplicitNetworks(networks []*networkTypes.Network) {
-	for _, network := range networks {
-		if network == nil {
-			continue
-		}
-		log.WithContext(r.ctx).Tracef("collectFromExplicitNetworks: changed network %s -> marking for bridge", network.ID)
-		if network.ID != "" {
-			r.networkIDs[network.ID] = struct{}{}
-		}
-	}
-}
-
-func (r *resolver) collectFromPostureChecks(postureCheckIDs []string) {
-	if len(postureCheckIDs) == 0 {
+// foldPolicySide folds one side of a policy down to affected peers: its groups
+// (resolved to members in expand) and its direct peer. When the side is the
+// DESTINATION and references a network resource (directly or via a destination
+// group's resources), it also folds the routers that serve that resource's network
+// — a destination resource is reached through its routers. A resource on the SOURCE
+// side routes to nobody (GetPoliciesForNetworkResource matches destinations only),
+// so the router hop is destination-only.
+func (r *resolver) foldPolicySide(policy *types.Policy, side policySide) {
+	if policy == nil {
 		return
 	}
+	for _, rule := range policy.Rules {
+		addAll(r.affectedGroups, ruleSideGroups(rule, side))
+		res := ruleSideResource(rule, side)
+		if res.Type == types.ResourceTypePeer && res.ID != "" {
+			r.affectedPeers[res.ID] = struct{}{}
+		}
+	}
+	if side == sideDestination {
+		r.foldRoutersForResources(r.policyDestinationResourceIDs(policy))
+	}
+}
+
+// appendPoliciesForPostureChecks appends every policy that references a changed
+// posture check (a rule change, so walk both sides).
+func (r *resolver) appendPoliciesForPostureChecks(policies []*types.Policy, postureCheckIDs []string) []*types.Policy {
+	if len(postureCheckIDs) == 0 {
+		return policies
+	}
 	ids := toSet(postureCheckIDs)
 	for _, policy := range r.policies() {
 		if !policyReferencesPostureChecks(policy, ids) {
 			continue
 		}
-		log.WithContext(r.ctx).Tracef("collectFromPostureChecks: policy %s (%s) references changed posture checks %v -> folding rule groups %v + direct peers",
-			policy.ID, policy.Name, postureCheckIDs, policy.RuleGroups())
-		addAll(r.groupSet, policy.RuleGroups())
-		collectPolicyDirectPeers(policy, r.peerSet)
-		r.matchedPolicies = append(r.matchedPolicies, policy)
+		log.WithContext(r.ctx).Tracef("appendPoliciesForPostureChecks: policy %s (%s) references changed posture checks %v -> both-sides policy",
+			policy.ID, policy.Name, postureCheckIDs)
+		policies = append(policies, policy)
+	}
+	return policies
+}
+
+// collectFromPolicies folds, for every policy whose rule a changed group or peer
+// touches, only the OPPOSITE side (down to peers, incl. destination routers), plus
+// the changed entity's own side: the changed group's whole membership when the
+// group itself changed (outputGroups), or the changed peer alone when matched via a
+// peer-seeded group (never its co-members).
+func (r *resolver) collectFromPolicies() {
+	for _, policy := range r.policies() {
+		for _, rule := range policy.Rules {
+			if !rule.Enabled {
+				continue // a disabled rule grants no access
+			}
+			r.foldRuleSideIfChanged(policy, rule, sideSource)
+			r.foldRuleSideIfChanged(policy, rule, sideDestination)
+		}
 	}
 }
 
-func (r *resolver) collectFromPolicies() {
-	for _, policy := range r.policies() {
-		matchedByGroup := policyReferencesGroups(policy, r.changedGroupSet)
-		matchedByPeer := len(r.changedPeerSet) > 0 && policyReferencesDirectPeers(policy, r.changedPeerSet)
-		if !matchedByGroup && !matchedByPeer {
+// foldRuleSideIfChanged: when a changed group or direct peer sits on `side` of the
+// rule, fold the opposite side fully (groups/peers + destination routers) and fold
+// the changed entity's own side (the whole changed group, or the changed peer alone).
+func (r *resolver) foldRuleSideIfChanged(policy *types.Policy, rule *types.PolicyRule, side policySide) {
+	nearGroups := ruleSideGroups(rule, side)
+	nearResource := ruleSideResource(rule, side)
+
+	matchedByGroup := anyInSet(nearGroups, r.linkGroups)
+	matchedByPeer := isDirectPeerInSet(nearResource, r.changedPeers)
+	if !matchedByGroup && !matchedByPeer {
+		return
+	}
+
+	// Opposite side, fully down to peers (a destination opposite also folds routers).
+	r.foldPolicySideForRule(policy, rule, side.opposite())
+
+	// Own side: fold the whole changed group's members only when the group itself
+	// changed (outputGroups). A peer-seeded or link-only group is not folded here —
+	// its siblings never refresh. The changed peers themselves are folded once, after
+	// the walk (see walk()).
+	for _, gID := range nearGroups {
+		if _, ok := r.outputGroups[gID]; ok {
+			r.affectedGroups[gID] = struct{}{}
+		}
+	}
+
+	// When the changed side IS a destination, the resources it targets are reached
+	// through their network's routers, so those routers refresh too (e.g. attaching a
+	// resource to a destination group, or a changed destination group/resource).
+	if side == sideDestination {
+		r.foldRoutersForResources(r.ruleDestinationResourceIDs(rule))
+	}
+}
+
+// foldPolicySideForRule folds one side of a single rule (groups + direct peer), and
+// for a destination side the routers of that rule's destination resources.
+func (r *resolver) foldPolicySideForRule(policy *types.Policy, rule *types.PolicyRule, side policySide) {
+	addAll(r.affectedGroups, ruleSideGroups(rule, side))
+	res := ruleSideResource(rule, side)
+	if res.Type == types.ResourceTypePeer && res.ID != "" {
+		r.affectedPeers[res.ID] = struct{}{}
+	}
+	if side == sideDestination {
+		r.foldRoutersForResources(r.ruleDestinationResourceIDs(rule))
+	}
+}
+
+// collectFromChangedRoutes folds an explicitly changed route's own groups and peer.
+func (r *resolver) collectFromChangedRoutes(routes []*route.Route) {
+	for _, rt := range routes {
+		if rt == nil {
 			continue
 		}
-		log.WithContext(r.ctx).Tracef("collectFromPolicies: policy %s (%s) matched (byGroup=%t byPeer=%t) -> folding rule groups %v + direct peers",
-			policy.ID, policy.Name, matchedByGroup, matchedByPeer, policy.RuleGroups())
-		addAll(r.groupSet, policy.RuleGroups())
-		collectPolicyDirectPeers(policy, r.peerSet)
-		r.matchedPolicies = append(r.matchedPolicies, policy)
+		log.WithContext(r.ctx).Tracef("collectFromChangedRoutes: changed route %s -> folding groups=%v peerGroups=%v accessControlGroups=%v peer=%q",
+			rt.ID, rt.Groups, rt.PeerGroups, rt.AccessControlGroups, rt.Peer)
+		addAll(r.affectedGroups, rt.Groups, rt.PeerGroups, rt.AccessControlGroups)
+		if rt.Peer != "" {
+			r.affectedPeers[rt.Peer] = struct{}{}
+		}
+	}
+}
+
+// collectFromChangedRouters: a changed router refreshes its OWN backing peer/groups
+// (the changed entity) and the SOURCE side of every policy reaching a resource on
+// its network (the router serves the whole network). Sibling routers on the network
+// are independent and are NOT folded. Passing the old router state keeps a repointed
+// router's previous backing affected without a post-commit read.
+func (r *resolver) collectFromChangedRouters(routers []*routerTypes.NetworkRouter) {
+	for _, router := range routers {
+		if router == nil {
+			continue
+		}
+		log.WithContext(r.ctx).Tracef("collectFromChangedRouters: changed router %s on network %s -> folding its own peerGroups=%v peer=%q + sources reaching network resources",
+			router.ID, router.NetworkID, router.PeerGroups, router.Peer)
+		addAll(r.affectedGroups, router.PeerGroups)
+		if router.Peer != "" {
+			r.affectedPeers[router.Peer] = struct{}{}
+		}
+		if router.NetworkID != "" {
+			r.foldPolicySourcesForResources(r.networkResourceIDs(router.NetworkID))
+		}
+	}
+}
+
+// collectFromChangedResources: a changed resource refreshes the SOURCE side of the
+// policies targeting EXACTLY that resource — directly, or via one of the resource's
+// own groups (old∪new across the change, so a now-detached group's sources still
+// refresh) — plus the routers serving its network (the resource is reached through
+// them). It does not touch sibling resources on the same network.
+func (r *resolver) collectFromChangedResources(resources []*resourceTypes.NetworkResource) {
+	for _, resource := range resources {
+		if resource == nil {
+			continue
+		}
+		log.WithContext(r.ctx).Tracef("collectFromChangedResources: changed resource %s on network %s (groups %v) -> folding sources of policies targeting it + its network's routers",
+			resource.ID, resource.NetworkID, resource.GroupIDs)
+		r.foldPolicySourcesForResource(resource.ID, resource.GroupIDs)
+		if resource.NetworkID != "" {
+			r.foldRoutersOnNetworks(map[string]struct{}{resource.NetworkID: {}})
+		}
+	}
+}
+
+// foldPolicySourcesForResource folds the source side of every policy whose
+// destination is the given resource — referenced directly, or via any of the given
+// groups (the resource's own old∪new groups, which captures a detached group).
+func (r *resolver) foldPolicySourcesForResource(resourceID string, groupIDs []string) {
+	groups := toSet(groupIDs)
+	for _, policy := range r.policies() {
+		if !policyTargetsResourceOrGroups(policy, resourceID, groups) {
+			continue
+		}
+		log.WithContext(r.ctx).Tracef("foldPolicySourcesForResource: policy %s (%s) targets changed resource %s -> folding its source groups/peers", policy.ID, policy.Name, resourceID)
+		collectPolicySources(policy, r.affectedGroups, r.affectedPeers)
+	}
+}
+
+// policyTargetsResourceOrGroups reports whether a policy's destination is the given
+// resource directly, or one of the given destination groups.
+func policyTargetsResourceOrGroups(policy *types.Policy, resourceID string, groups map[string]struct{}) bool {
+	if policy == nil {
+		return false
+	}
+	for _, rule := range policy.Rules {
+		if !rule.Enabled {
+			continue
+		}
+		if rule.DestinationResource.Type != types.ResourceTypePeer && rule.DestinationResource.ID == resourceID && resourceID != "" {
+			return true
+		}
+		if anyInSet(rule.Destinations, groups) {
+			return true
+		}
+	}
+	return false
+}
+
+// collectFromChangedNetworks: a changed network refreshes the SOURCE side of the
+// policies reaching any of its resources, plus its routers. A network has no
+// groups/peers of its own.
+func (r *resolver) collectFromChangedNetworks(networks []*networkTypes.Network) {
+	for _, network := range networks {
+		if network == nil || network.ID == "" {
+			continue
+		}
+		log.WithContext(r.ctx).Tracef("collectFromChangedNetworks: changed network %s -> folding sources reaching its resources + its routers", network.ID)
+		resourceIDs := r.networkResourceIDs(network.ID)
+		r.foldPolicySourcesForResources(resourceIDs)
+		r.foldRoutersOnNetworks(map[string]struct{}{network.ID: {}})
+	}
+}
+
+// foldPolicySourcesForResources folds the source groups/peers of every policy whose
+// destination targets one of resourceIDs (directly or via a destination group).
+func (r *resolver) foldPolicySourcesForResources(resourceIDs map[string]struct{}) {
+	if len(resourceIDs) == 0 {
+		return
+	}
+	for _, policy := range r.policies() {
+		if r.policyTargetsResources(policy, resourceIDs) {
+			log.WithContext(r.ctx).Tracef("foldPolicySourcesForResources: policy %s (%s) targets a changed resource -> folding its source groups/peers", policy.ID, policy.Name)
+			collectPolicySources(policy, r.affectedGroups, r.affectedPeers)
+		}
 	}
 }
 
 func (r *resolver) collectFromRoutes() {
 	for _, rt := range r.snap.routes {
-		matchedByGroup := anyInSet(rt.Groups, r.changedGroupSet) || anyInSet(rt.PeerGroups, r.changedGroupSet) || anyInSet(rt.AccessControlGroups, r.changedGroupSet)
-		matchedByPeer := rt.Peer != "" && len(r.changedPeerSet) > 0 && isInSet(rt.Peer, r.changedPeerSet)
+		if !rt.Enabled {
+			continue // disabled routes route to nobody; skip existing account data
+		}
+		matchedByGroup := anyInSet(rt.Groups, r.linkGroups) || anyInSet(rt.PeerGroups, r.linkGroups) || anyInSet(rt.AccessControlGroups, r.linkGroups)
+		matchedByPeer := rt.Peer != "" && len(r.changedPeers) > 0 && isInSet(rt.Peer, r.changedPeers)
 		if !matchedByGroup && !matchedByPeer {
 			continue
 		}
 		log.WithContext(r.ctx).Tracef("collectFromRoutes: route %s matched (byGroup=%t byPeer=%t) -> folding groups=%v peerGroups=%v accessControlGroups=%v peer=%q",
 			rt.ID, matchedByGroup, matchedByPeer, rt.Groups, rt.PeerGroups, rt.AccessControlGroups, rt.Peer)
-		addAll(r.groupSet, rt.Groups, rt.PeerGroups, rt.AccessControlGroups)
+		addAll(r.affectedGroups, rt.Groups, rt.PeerGroups, rt.AccessControlGroups)
 		if rt.Peer != "" {
-			r.peerSet[rt.Peer] = struct{}{}
+			r.affectedPeers[rt.Peer] = struct{}{}
 		}
 	}
 }
 
 func (r *resolver) collectFromNameServers() {
-	if len(r.changedGroupSet) == 0 {
+	if len(r.linkGroups) == 0 {
 		return
 	}
 	for _, ns := range r.snap.nsGroups {
-		if anyInSet(ns.Groups, r.changedGroupSet) {
+		if anyInSet(ns.Groups, r.linkGroups) {
 			log.WithContext(r.ctx).Tracef("collectFromNameServers: nameserver group %s references a changed group -> folding its groups %v", ns.ID, ns.Groups)
-			addAll(r.groupSet, ns.Groups)
+			addAll(r.affectedGroups, ns.Groups)
 		}
 	}
 }
 
 func (r *resolver) collectFromDNSSettings() {
-	if len(r.changedGroupSet) == 0 || r.snap.dnsSettings == nil {
+	if len(r.linkGroups) == 0 || r.snap.dnsSettings == nil {
 		return
 	}
 	for _, gID := range r.snap.dnsSettings.DisabledManagementGroups {
-		if _, ok := r.changedGroupSet[gID]; ok {
+		if _, ok := r.linkGroups[gID]; ok {
 			log.WithContext(r.ctx).Tracef("collectFromDNSSettings: changed group %s is in DisabledManagementGroups -> folding it", gID)
-			r.groupSet[gID] = struct{}{}
+			r.affectedGroups[gID] = struct{}{}
 		}
 	}
 }
 
+// collectFromNetworkRouters handles a changed group/peer that BACKS a router (the
+// routing peer set moved): the router's own peers refresh and so do the sources of
+// the policies reaching its network's resources. Sibling routers on the network are
+// independent and are not folded.
 func (r *resolver) collectFromNetworkRouters() {
 	for _, router := range r.networkRouters() {
-		matchedByGroup := anyInSet(router.PeerGroups, r.changedGroupSet)
-		matchedByPeer := router.Peer != "" && len(r.changedPeerSet) > 0 && isInSet(router.Peer, r.changedPeerSet)
+		matchedByGroup := anyInSet(router.PeerGroups, r.linkGroups)
+		matchedByPeer := router.Peer != "" && len(r.changedPeers) > 0 && isInSet(router.Peer, r.changedPeers)
 		if !matchedByGroup && !matchedByPeer {
 			continue
 		}
-		log.WithContext(r.ctx).Tracef("collectFromNetworkRouters: router %s on network %s matched (byGroup=%t byPeer=%t) -> folding peerGroups=%v peer=%q and marking network for source bridge",
+		log.WithContext(r.ctx).Tracef("collectFromNetworkRouters: router %s on network %s matched (byGroup=%t byPeer=%t) -> folding its peerGroups=%v peer=%q + sources reaching network resources",
 			router.ID, router.NetworkID, matchedByGroup, matchedByPeer, router.PeerGroups, router.Peer)
-		addAll(r.groupSet, router.PeerGroups)
+		addAll(r.affectedGroups, router.PeerGroups)
 		if router.Peer != "" {
-			r.peerSet[router.Peer] = struct{}{}
+			r.affectedPeers[router.Peer] = struct{}{}
+		}
+		if router.NetworkID != "" {
+			r.foldPolicySourcesForResources(r.networkResourceIDs(router.NetworkID))
 		}
-		r.networkIDs[router.NetworkID] = struct{}{}
 	}
 }
 
@@ -526,42 +787,45 @@ func (r *resolver) collectFromProxyServices() {
 	expanded := r.expandChangedPeersWithGroups()
 
 	for _, svc := range services {
-		if svc == nil {
-			continue
+		if svc == nil || !svc.Enabled {
+			continue // a disabled service proxies nothing; skip existing account data
 		}
 		proxyPeers := proxyByCluster[svc.ProxyCluster]
 		if len(proxyPeers) == 0 {
 			continue
 		}
 		matchedByPeer := serviceMatchesChangedPeers(svc, proxyPeers, expanded)
-		matchedByAccessGroup := anyInSet(svc.AccessGroups, r.changedGroupSet)
+		matchedByAccessGroup := anyInSet(svc.AccessGroups, r.linkGroups)
 		if !matchedByPeer && !matchedByAccessGroup {
 			continue
 		}
 		log.WithContext(r.ctx).Tracef("collectFromProxyServices: service %s (cluster=%s) matched (byProxyOrTargetPeer=%t byAccessGroup=%t) -> folding %d proxy peers, peer targets and access groups %v",
 			svc.ID, svc.ProxyCluster, matchedByPeer, matchedByAccessGroup, len(proxyPeers), svc.AccessGroups)
 		for _, pid := range proxyPeers {
-			r.peerSet[pid] = struct{}{}
+			r.affectedPeers[pid] = struct{}{}
 		}
 		for _, target := range svc.Targets {
+			if !target.Enabled {
+				continue // a disabled target forwards nothing
+			}
 			if target.TargetType == rpservice.TargetTypePeer && target.TargetId != "" {
-				r.peerSet[target.TargetId] = struct{}{}
+				r.affectedPeers[target.TargetId] = struct{}{}
 			}
 		}
-		addAll(r.groupSet, svc.AccessGroups)
+		addAll(r.affectedGroups, svc.AccessGroups)
 	}
 }
 
 func (r *resolver) expandChangedPeersWithGroups() map[string]struct{} {
-	if len(r.changedGroupSet) == 0 {
-		return r.changedPeerSet
+	if len(r.linkGroups) == 0 {
+		return r.changedPeers
 	}
-	ids := r.peerIDsForGroups(r.changedGroupSet)
+	ids := r.peerIDsForGroups(r.linkGroups)
 	if len(ids) == 0 {
-		return r.changedPeerSet
+		return r.changedPeers
 	}
-	merged := make(map[string]struct{}, len(r.changedPeerSet)+len(ids))
-	for id := range r.changedPeerSet {
+	merged := make(map[string]struct{}, len(r.changedPeers)+len(ids))
+	for id := range r.changedPeers {
 		merged[id] = struct{}{}
 	}
 	for _, id := range ids {
@@ -570,54 +834,36 @@ func (r *resolver) expandChangedPeersWithGroups() map[string]struct{} {
 	return merged
 }
 
-// collectResourceRouterBridge crosses between source peers and routing peers, which
-// are reachable only via resource -> network -> router, not through the policy's own
-// groups: source -> router (targeted resources' networks), then router -> source.
-func (r *resolver) collectResourceRouterBridge() {
-	r.bridgeSourceToRouters()
-	r.bridgeRoutersToSources()
-}
-
-func (r *resolver) bridgeSourceToRouters() {
-	resourceIDs := r.policyDestinationResourceIDs(r.matchedPolicies...)
+// foldRoutersForResources folds the routers serving the networks of the given
+// resources (a destination resource is reached through its network's routers). It is
+// the resource -> network -> router hop used by foldPolicySide for a destination.
+func (r *resolver) foldRoutersForResources(resourceIDs map[string]struct{}) {
 	if len(resourceIDs) == 0 {
 		return
 	}
-
-	networkIDs := r.resourceNetworkIDs(resourceIDs)
-	log.WithContext(r.ctx).Tracef("bridgeSourceToRouters: targeted resources %v -> networks %v (their routers become affected via the router->source pass)",
-		setToSlice(resourceIDs), setToSlice(networkIDs))
-	for id := range networkIDs {
-		r.networkIDs[id] = struct{}{}
-	}
+	r.foldRoutersOnNetworks(r.resourceNetworkIDs(resourceIDs))
 }
 
-func (r *resolver) bridgeRoutersToSources() {
-	if len(r.networkIDs) == 0 {
-		return
+// ruleDestinationResourceIDs returns the destination resource IDs of a single rule:
+// the direct DestinationResource plus the resources of its destination groups.
+func (r *resolver) ruleDestinationResourceIDs(rule *types.PolicyRule) map[string]struct{} {
+	resourceIDs := make(map[string]struct{})
+	if rule.DestinationResource.Type != types.ResourceTypePeer && rule.DestinationResource.ID != "" {
+		resourceIDs[rule.DestinationResource.ID] = struct{}{}
 	}
+	r.addGroupResourceIDs(toSet(rule.Destinations), resourceIDs)
+	return resourceIDs
+}
 
-	log.WithContext(r.ctx).Tracef("bridgeRoutersToSources: affected networks %v -> folding their routing peers and the source peers of policies targeting their resources",
-		setToSlice(r.networkIDs))
-
-	r.foldRoutersOnNetworks(r.networkIDs)
-
+// networkResourceIDs returns the IDs of all resources on the given network.
+func (r *resolver) networkResourceIDs(networkID string) map[string]struct{} {
 	resourceIDs := make(map[string]struct{})
 	for _, resource := range r.networkResources() {
-		if _, ok := r.networkIDs[resource.NetworkID]; ok {
+		if resource.NetworkID == networkID {
 			resourceIDs[resource.ID] = struct{}{}
 		}
 	}
-	if len(resourceIDs) == 0 {
-		return
-	}
-
-	for _, policy := range r.policies() {
-		if r.policyTargetsResources(policy, resourceIDs) {
-			log.WithContext(r.ctx).Tracef("bridgeRoutersToSources: policy %s (%s) targets an affected-network resource -> folding its source groups/peers", policy.ID, policy.Name)
-			collectPolicySources(policy, r.groupSet, r.peerSet)
-		}
-	}
+	return resourceIDs
 }
 
 func (r *resolver) foldRoutersOnNetworks(networkIDs map[string]struct{}) {
@@ -627,9 +873,9 @@ func (r *resolver) foldRoutersOnNetworks(networkIDs map[string]struct{}) {
 		}
 		log.WithContext(r.ctx).Tracef("bridgeRoutersToSources: router %s serves affected network %s -> folding peerGroups=%v peer=%q",
 			router.ID, router.NetworkID, router.PeerGroups, router.Peer)
-		addAll(r.groupSet, router.PeerGroups)
+		addAll(r.affectedGroups, router.PeerGroups)
 		if router.Peer != "" {
-			r.peerSet[router.Peer] = struct{}{}
+			r.affectedPeers[router.Peer] = struct{}{}
 		}
 	}
 }
@@ -650,6 +896,9 @@ func (r *resolver) policyTargetsResources(policy *types.Policy, resourceIDs map[
 	}
 	destGroupSet := make(map[string]struct{})
 	for _, rule := range policy.Rules {
+		if !rule.Enabled {
+			continue
+		}
 		if rule.DestinationResource.Type != types.ResourceTypePeer && isInSet(rule.DestinationResource.ID, resourceIDs) {
 			return true
 		}
@@ -714,44 +963,20 @@ func (r *resolver) addGroupResourceIDs(groupIDs map[string]struct{}, resourceIDs
 	}
 }
 
-func collectPolicyDirectPeers(policy *types.Policy, peerSet map[string]struct{}) {
+// collectPolicySources folds the source groups/peers of a snapshot policy's enabled
+// rules (a disabled rule grants no access).
+func collectPolicySources(policy *types.Policy, groups, peers map[string]struct{}) {
 	for _, rule := range policy.Rules {
+		if !rule.Enabled {
+			continue
+		}
+		addAll(groups, rule.Sources)
 		if rule.SourceResource.Type == types.ResourceTypePeer && rule.SourceResource.ID != "" {
-			peerSet[rule.SourceResource.ID] = struct{}{}
-		}
-		if rule.DestinationResource.Type == types.ResourceTypePeer && rule.DestinationResource.ID != "" {
-			peerSet[rule.DestinationResource.ID] = struct{}{}
+			peers[rule.SourceResource.ID] = struct{}{}
 		}
 	}
 }
 
-func collectPolicySources(policy *types.Policy, groupSet, peerSet map[string]struct{}) {
-	for _, rule := range policy.Rules {
-		addAll(groupSet, rule.Sources)
-		if rule.SourceResource.Type == types.ResourceTypePeer && rule.SourceResource.ID != "" {
-			peerSet[rule.SourceResource.ID] = struct{}{}
-		}
-	}
-}
-
-func policyReferencesGroups(policy *types.Policy, groupSet map[string]struct{}) bool {
-	for _, rule := range policy.Rules {
-		if anyInSet(rule.Sources, groupSet) || anyInSet(rule.Destinations, groupSet) {
-			return true
-		}
-	}
-	return false
-}
-
-func policyReferencesDirectPeers(policy *types.Policy, changedSet map[string]struct{}) bool {
-	for _, rule := range policy.Rules {
-		if isDirectPeerInSet(rule.SourceResource, changedSet) || isDirectPeerInSet(rule.DestinationResource, changedSet) {
-			return true
-		}
-	}
-	return false
-}
-
 func policyReferencesPostureChecks(policy *types.Policy, ids map[string]struct{}) bool {
 	for _, id := range policy.SourcePostureChecks {
 		if _, ok := ids[id]; ok {
@@ -776,7 +1001,7 @@ func serviceMatchesChangedPeers(svc *rpservice.Service, proxyPeers []string, cha
 		}
 	}
 	for _, target := range svc.Targets {
-		if target.TargetType != rpservice.TargetTypePeer || target.TargetId == "" {
+		if !target.Enabled || target.TargetType != rpservice.TargetTypePeer || target.TargetId == "" {
 			continue
 		}
 		if _, ok := changedPeers[target.TargetId]; ok {

management/server/affectedpeers/resolver_test.go

@@ -10,8 +10,8 @@ import (
 	"github.com/netbirdio/netbird/management/server/types"
 )
 
-// policyGroupsAndPeers mirrors the explicit-policy extraction (RuleGroups +
-// direct peers) the resolver folds in, for asserting the pure logic.
+// policyGroupsAndPeers mirrors the both-sides extraction (RuleGroups + direct peers)
+// the resolver folds in for a changed policy, for asserting the pure logic.
 func policyGroupsAndPeers(policies ...*types.Policy) (groups []string, peers []string) {
 	peerSet := map[string]struct{}{}
 	for _, p := range policies {
@@ -19,7 +19,14 @@ func policyGroupsAndPeers(policies ...*types.Policy) (groups []string, peers []s
 			continue
 		}
 		groups = append(groups, p.RuleGroups()...)
-		collectPolicyDirectPeers(p, peerSet)
+		for _, rule := range p.Rules {
+			if rule.SourceResource.Type == types.ResourceTypePeer && rule.SourceResource.ID != "" {
+				peerSet[rule.SourceResource.ID] = struct{}{}
+			}
+			if rule.DestinationResource.Type == types.ResourceTypePeer && rule.DestinationResource.ID != "" {
+				peerSet[rule.DestinationResource.ID] = struct{}{}
+			}
+		}
 	}
 	for id := range peerSet {
 		peers = append(peers, id)
@@ -80,26 +87,6 @@ func TestChangeIsEmpty(t *testing.T) {
 	assert.False(t, Change{PostureCheckIDs: []string{"pc"}}.isEmpty())
 }
 
-func TestPolicyReferencesGroups(t *testing.T) {
-	policy := &types.Policy{Rules: []*types.PolicyRule{{Sources: []string{"g1", "g2"}, Destinations: []string{"g3"}}}}
-
-	assert.True(t, policyReferencesGroups(policy, map[string]struct{}{"g1": {}}))
-	assert.True(t, policyReferencesGroups(policy, map[string]struct{}{"g3": {}}))
-	assert.False(t, policyReferencesGroups(policy, map[string]struct{}{"g4": {}}))
-	assert.False(t, policyReferencesGroups(policy, map[string]struct{}{}))
-}
-
-func TestPolicyReferencesDirectPeers(t *testing.T) {
-	policy := &types.Policy{Rules: []*types.PolicyRule{{
-		SourceResource:      types.Resource{Type: types.ResourceTypePeer, ID: "p1"},
-		DestinationResource: types.Resource{Type: types.ResourceTypeHost, ID: "r1"},
-	}}}
-
-	assert.True(t, policyReferencesDirectPeers(policy, map[string]struct{}{"p1": {}}))
-	assert.False(t, policyReferencesDirectPeers(policy, map[string]struct{}{"r1": {}}))
-	assert.False(t, policyReferencesDirectPeers(policy, map[string]struct{}{"p2": {}}))
-}
-
 func TestPolicyReferencesPostureChecks(t *testing.T) {
 	policy := &types.Policy{SourcePostureChecks: []string{"pc1", "pc2"}}
 
@@ -107,24 +94,9 @@ func TestPolicyReferencesPostureChecks(t *testing.T) {
 	assert.False(t, policyReferencesPostureChecks(policy, map[string]struct{}{"pc3": {}}))
 }
 
-func TestCollectPolicyDirectPeers(t *testing.T) {
-	policy := &types.Policy{Rules: []*types.PolicyRule{{
-		SourceResource:      types.Resource{Type: types.ResourceTypePeer, ID: "p1"},
-		DestinationResource: types.Resource{Type: types.ResourceTypePeer, ID: "p2"},
-	}, {
-		DestinationResource: types.Resource{Type: types.ResourceTypeHost, ID: "r1"},
-	}}}
-
-	peerSet := map[string]struct{}{}
-	collectPolicyDirectPeers(policy, peerSet)
-
-	assert.Contains(t, peerSet, "p1")
-	assert.Contains(t, peerSet, "p2")
-	assert.NotContains(t, peerSet, "r1")
-}
-
 func TestCollectPolicySources(t *testing.T) {
 	policy := &types.Policy{Rules: []*types.PolicyRule{{
+		Enabled:        true,
 		Sources:        []string{"g1"},
 		SourceResource: types.Resource{Type: types.ResourceTypePeer, ID: "p1"},
 		Destinations:   []string{"g2"},

management/server/group.go

@@ -520,7 +520,12 @@ func collectDeletableGroups(ctx context.Context, transaction store.Store, accoun
 // GroupAddPeer appends peer to the group
 func (am *DefaultAccountManager) GroupAddPeer(ctx context.Context, accountID, groupID, peerID string) error {
 	var snap *affectedpeers.Snapshot
-	change := affectedpeers.Change{ChangedGroupIDs: []string{groupID}}
+	// A membership change affects only the peer itself and the opposite side of THIS
+	// group's policies — not the group's other members, and not the peer's other
+	// groups. LinkGroups walks only this group (matched, not expanded); OutputPeerIDs
+	// refreshes the peer without seeding its other group memberships. For an
+	// intra-group policy the opposite side is the group, so its members still refresh.
+	change := affectedpeers.Change{OutputPeerIDs: []string{peerID}, LinkGroups: []string{groupID}}
 
 	err := am.Store.ExecuteInTransaction(ctx, func(transaction store.Store) error {
 		if err := transaction.AddPeerToGroup(ctx, accountID, peerID, groupID); err != nil {
@@ -586,10 +591,11 @@ func (am *DefaultAccountManager) GroupAddResource(ctx context.Context, accountID
 // GroupDeletePeer removes peer from the group
 func (am *DefaultAccountManager) GroupDeletePeer(ctx context.Context, accountID, groupID, peerID string) error {
 	var snap *affectedpeers.Snapshot
-	change := affectedpeers.Change{
-		ChangedGroupIDs:     []string{groupID},
-		RemovedPeersByGroup: map[string][]string{groupID: {peerID}},
-	}
+	// Same as GroupAddPeer: the removed peer and the opposite side of THIS group's
+	// policies refresh, not the group's other members or the peer's other groups. The
+	// peer is no longer in the group's index, but LinkGroups still drives the
+	// opposite-side walk, and OutputPeerIDs refreshes the removed peer itself.
+	change := affectedpeers.Change{OutputPeerIDs: []string{peerID}, LinkGroups: []string{groupID}}
 
 	err := am.Store.ExecuteInTransaction(ctx, func(transaction store.Store) error {
 		if err := transaction.RemovePeerFromGroup(ctx, peerID, groupID); err != nil {
@@ -600,8 +606,6 @@ func (am *DefaultAccountManager) GroupDeletePeer(ctx context.Context, accountID,
 			return err
 		}
 
-		// The removed peer is carried in change.RemovedPeersByGroup and folded in
-		// only when the group is linked, so loading post-removal is correct.
 		var err error
 		if snap, err = affectedpeers.Load(ctx, transaction, accountID, change); err != nil {
 			return err

management/server/http/handlers/users/users_handler.go

@@ -220,7 +220,7 @@ func (h *handler) getAllUsers(w http.ResponseWriter, r *http.Request) {
 		}
 
 		includeServiceUser, err := strconv.ParseBool(serviceUser)
-		log.WithContext(r.Context()).Debugf("Should include service user: %v", includeServiceUser)
+		log.WithContext(r.Context()).Tracef("Should include service user: %v", includeServiceUser)
 		if err != nil {
 			util.WriteError(r.Context(), status.Errorf(status.InvalidArgument, "invalid service_user query parameter"), w)
 			return

management/server/mock_server/account_mock.go

@@ -39,7 +39,7 @@ type MockAccountManager struct {
 	GetUserFromUserAuthFunc               func(ctx context.Context, userAuth auth.UserAuth) (*types.User, error)
 	ListUsersFunc                         func(ctx context.Context, accountID string) ([]*types.User, error)
 	GetPeersFunc                          func(ctx context.Context, accountID, userID, nameFilter, ipFilter string) ([]*nbpeer.Peer, error)
-	MarkPeerConnectedFunc                 func(ctx context.Context, peerKey string, realIP net.IP, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error
+	MarkPeerConnectedFunc                 func(ctx context.Context, peerKey string, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error
 	MarkPeerDisconnectedFunc              func(ctx context.Context, peerKey string, accountID string, sessionStartedAt int64) error
 	SyncAndMarkPeerFunc                   func(ctx context.Context, accountID string, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP, syncTime time.Time) (*nbpeer.Peer, *types.NetworkMap, []*posture.Checks, int64, error)
 	DeletePeerFunc                        func(ctx context.Context, accountID, peerKey, userID string) error
@@ -114,7 +114,7 @@ type MockAccountManager struct {
 	GetIdpManagerFunc                     func() idp.Manager
 	UpdateIntegratedValidatorFunc         func(ctx context.Context, accountID, userID, validator string, groups []string) error
 	GroupValidationFunc                   func(ctx context.Context, accountId string, groups []string) (bool, error)
-	SyncPeerMetaFunc                      func(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta) error
+	SyncPeerMetaFunc                      func(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP) error
 	FindExistingPostureCheckFunc          func(accountID string, checks *posture.ChecksDefinition) (*posture.Checks, error)
 	GetAccountIDForPeerKeyFunc            func(ctx context.Context, peerKey string) (string, error)
 	GetAccountByIDFunc                    func(ctx context.Context, accountID string, userID string) (*types.Account, error)
@@ -345,9 +345,9 @@ func (am *MockAccountManager) GetAccountIDByUserID(ctx context.Context, userAuth
 }
 
 // MarkPeerConnected mock implementation of MarkPeerConnected from server.AccountManager interface
-func (am *MockAccountManager) MarkPeerConnected(ctx context.Context, peerKey string, realIP net.IP, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error {
+func (am *MockAccountManager) MarkPeerConnected(ctx context.Context, peerKey string, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error {
 	if am.MarkPeerConnectedFunc != nil {
-		return am.MarkPeerConnectedFunc(ctx, peerKey, realIP, accountID, sessionStartedAt, nmap)
+		return am.MarkPeerConnectedFunc(ctx, peerKey, accountID, sessionStartedAt, nmap)
 	}
 	return status.Errorf(codes.Unimplemented, "method MarkPeerConnected is not implemented")
 }
@@ -975,9 +975,9 @@ func (am *MockAccountManager) GroupValidation(ctx context.Context, accountId str
 }
 
 // SyncPeerMeta mocks SyncPeerMeta of the AccountManager interface
-func (am *MockAccountManager) SyncPeerMeta(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta) error {
+func (am *MockAccountManager) SyncPeerMeta(ctx context.Context, peerPubKey string, meta nbpeer.PeerSystemMeta, realIP net.IP) error {
 	if am.SyncPeerMetaFunc != nil {
-		return am.SyncPeerMetaFunc(ctx, peerPubKey, meta)
+		return am.SyncPeerMetaFunc(ctx, peerPubKey, meta, realIP)
 	}
 	return status.Errorf(codes.Unimplemented, "method SyncPeerMeta is not implemented")
 }

management/server/peer.go

@@ -74,7 +74,7 @@ func (am *DefaultAccountManager) GetPeers(ctx context.Context, accountID, userID
 //
 // Disconnects use MarkPeerDisconnected and require the session to match
 // exactly; see PeerStatus.SessionStartedAt for the protocol.
-func (am *DefaultAccountManager) MarkPeerConnected(ctx context.Context, peerPubKey string, realIP net.IP, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error {
+func (am *DefaultAccountManager) MarkPeerConnected(ctx context.Context, peerPubKey string, accountID string, sessionStartedAt int64, nmap *types.NetworkMap) error {
 	start := time.Now()
 	defer func() {
 		am.metrics.AccountManagerMetrics().RecordPeerStatusUpdateDuration(telemetry.PeerStatusConnect, time.Since(start))
@@ -102,10 +102,6 @@ func (am *DefaultAccountManager) MarkPeerConnected(ctx context.Context, peerPubK
 	}
 	am.metrics.AccountManagerMetrics().CountPeerStatusUpdate(telemetry.PeerStatusConnect, telemetry.PeerStatusApplied)
 
-	if am.geo != nil && realIP != nil {
-		am.updatePeerLocationIfChanged(ctx, accountID, peer, realIP)
-	}
-
 	if err = am.schedulePeerExpirations(ctx, accountID, peer); err != nil {
 		return err
 	}
@@ -192,27 +188,40 @@ func (am *DefaultAccountManager) MarkPeerDisconnected(ctx context.Context, peerP
 		}
 	}
 
+	if peer.AddedWithSSOLogin() && peer.InactivityExpirationEnabled {
+		settings, err := am.Store.GetAccountSettings(ctx, store.LockingStrengthNone, accountID)
+		if err != nil {
+			log.WithContext(ctx).Warnf("failed getting account settings to schedule inactivity expiration for peer %s: %v", peer.ID, err)
+		} else if settings.PeerInactivityExpirationEnabled {
+			am.checkAndSchedulePeerInactivityExpiration(ctx, accountID)
+		}
+	}
+
 	return nil
 }
 
-// updatePeerLocationIfChanged refreshes the geolocation on a separate
-// row update, only when the connection IP actually changed. Geo lookups
-// are expensive so we skip same-IP reconnects.
-func (am *DefaultAccountManager) updatePeerLocationIfChanged(ctx context.Context, accountID string, peer *nbpeer.Peer, realIP net.IP) {
+// resolvePeerLocation looks up the geo location for realIP, returning nil when
+// there is nothing to apply: geo disabled, no real IP, the IP is unchanged from
+// what the peer already has, or the lookup failed. Geo lookups are skipped on
+// same-IP reconnects since they are comparatively expensive. The returned value
+// is applied by Peer.UpdateMetaIfNew so the change is persisted by its peer save.
+func (am *DefaultAccountManager) resolvePeerLocation(ctx context.Context, peer *nbpeer.Peer, realIP net.IP) *nbpeer.Location {
+	if am.geo == nil || realIP == nil {
+		return nil
+	}
 	if peer.Location.ConnectionIP != nil && peer.Location.ConnectionIP.Equal(realIP) {
-		return
+		return nil
 	}
 	location, err := am.geo.Lookup(realIP)
 	if err != nil {
 		log.WithContext(ctx).Warnf("failed to get location for peer %s realip: [%s]: %v", peer.ID, realIP.String(), err)
-		return
+		return nil
 	}
-	peer.Location.ConnectionIP = realIP
-	peer.Location.CountryCode = location.Country.ISOCode
-	peer.Location.CityName = location.City.Names.En
-	peer.Location.GeoNameID = location.City.GeonameID
-	if err := am.Store.SavePeerLocation(ctx, accountID, peer); err != nil {
-		log.WithContext(ctx).Warnf("could not store location for peer %s: %s", peer.ID, err)
+	return &nbpeer.Location{
+		ConnectionIP: realIP,
+		CountryCode:  location.Country.ISOCode,
+		CityName:     location.City.Names.En,
+		GeoNameID:    location.City.GeonameID,
 	}
 }
 
@@ -721,7 +730,7 @@ func (am *DefaultAccountManager) handleSetupKeyAddedPeer(ctx context.Context, en
 func (am *DefaultAccountManager) AddPeer(ctx context.Context, accountID, setupKey, userID string, peer *nbpeer.Peer, temporary bool) (*nbpeer.Peer, *types.Network, []*posture.Checks, bool, error) {
 	if setupKey == "" && userID == "" && !peer.ProxyMeta.Embedded {
 		// no auth method provided => reject access
-		return nil, nil, nil, false, status.Errorf(status.Unauthenticated, "no peer auth method provided, please use a setup key or interactive SSO login")
+		return nil, nil, nil, false, status.ErrNoAuthMethodProvided
 	}
 
 	upperKey := strings.ToUpper(setupKey)
@@ -980,7 +989,8 @@ func getPeerIPDNSLabel(ip netip.Addr, peerHostName string) (string, error) {
 // SyncPeer checks whether peer is eligible for receiving NetworkMap (authenticated) and returns its NetworkMap if eligible
 func (am *DefaultAccountManager) SyncPeer(ctx context.Context, sync types.PeerSync, accountID string) (*nbpeer.Peer, *types.NetworkMap, []*posture.Checks, int64, error) {
 	var peer *nbpeer.Peer
-	var updated, versionChanged, ipv6CapabilityChanged bool
+	var ipv6CapabilityChanged bool
+	var metaDiff nbpeer.MetaDiff
 	var err error
 
 	settings, err := am.Store.GetAccountSettings(ctx, store.LockingStrengthNone, accountID)
@@ -1010,9 +1020,10 @@ func (am *DefaultAccountManager) SyncPeer(ctx context.Context, sync types.PeerSy
 		}
 
 		oldHasIPv6Cap := peer.HasCapability(nbpeer.PeerCapabilityIPv6Overlay)
-		updated, versionChanged = peer.UpdateMetaIfNew(ctx, sync.Meta)
+		newLocation := am.resolvePeerLocation(ctx, peer, sync.RealIP)
+		metaDiff = peer.UpdateMetaIfNew(ctx, sync.Meta, newLocation)
 		ipv6CapabilityChanged = oldHasIPv6Cap != peer.HasCapability(nbpeer.PeerCapabilityIPv6Overlay)
-		if updated {
+		if metaDiff.Updated() {
 			am.metrics.AccountManagerMetrics().CountPeerMetUpdate()
 			log.WithContext(ctx).Tracef("peer %s metadata updated", peer.ID)
 			if err = transaction.SavePeer(ctx, accountID, peer); err != nil {
@@ -1040,9 +1051,10 @@ func (am *DefaultAccountManager) SyncPeer(ctx context.Context, sync types.PeerSy
 		return nil, nil, nil, 0, err
 	}
 
-	if isStatusChanged || sync.UpdateAccountPeers || ipv6CapabilityChanged || (updated && (len(resPostureChecks) > 0 || versionChanged)) {
+	metaDiffAffectsPosture := posture.AffectsPosture(ctx, &metaDiff, resPostureChecks)
+	if isStatusChanged || sync.UpdateAccountPeers || ipv6CapabilityChanged || metaDiffAffectsPosture || metaDiff.VersionChanged() || metaDiff.HostnameChanged() {
 		changedPeerIDs := []string{peer.ID}
-		affectedPeerIDs := am.syncPeerAffectedPeers(ctx, accountID, peer.ID, nmap, peerNotValid, updated, len(resPostureChecks) > 0)
+		affectedPeerIDs := am.syncPeerAffectedPeers(ctx, accountID, peer.ID, nmap, peerNotValid, metaDiffAffectsPosture)
 		if err = am.networkMapController.OnPeersUpdated(ctx, accountID, changedPeerIDs, affectedPeerIDs); err != nil {
 			return nil, nil, nil, 0, fmt.Errorf("notify network map controller of peer update: %w", err)
 		}
@@ -1059,8 +1071,8 @@ func (am *DefaultAccountManager) SyncPeer(ctx context.Context, sync types.PeerSy
 // metadata change that flips a posture result removes this peer from others'
 // maps asymmetrically; that case (and an invalid peer, whose map is empty) falls
 // back to the resolver.
-func (am *DefaultAccountManager) syncPeerAffectedPeers(ctx context.Context, accountID, peerID string, nmap *types.NetworkMap, peerNotValid, metaUpdated, hasPostureChecks bool) []string {
-	if peerNotValid || (metaUpdated && hasPostureChecks) {
+func (am *DefaultAccountManager) syncPeerAffectedPeers(ctx context.Context, accountID, peerID string, nmap *types.NetworkMap, peerNotValid, metaChangeAffectedPosture bool) []string {
+	if peerNotValid || metaChangeAffectedPosture {
 		return am.resolveAffectedPeersForPeerChanges(ctx, am.Store, accountID, []string{peerID})
 	}
 	return affectedPeerIDsFromNetworkMap(nmap, peerID)
@@ -1170,7 +1182,7 @@ func (am *DefaultAccountManager) LoginPeer(ctx context.Context, login types.Peer
 	}
 
 	// This is needed to keep in memory for the peer config. Otherwise browser client will end in a retry loop
-	peer.UpdateMetaIfNew(ctx, login.Meta)
+	peer.Meta = login.Meta
 
 	peerGroupIDs, err = getPeerGroupIDs(ctx, am.Store, accountID, peer.ID)
 	if err != nil {

management/server/peer/peer.go

@@ -107,6 +107,15 @@ type Location struct {
 	GeoNameID    uint // city level geoname id
 }
 
+// equal reports whether two locations match. ConnectionIP is a net.IP slice, so it uses
+// IP.Equal, not ==.
+func (l Location) equal(other Location) bool {
+	return l.CountryCode == other.CountryCode &&
+		l.CityName == other.CityName &&
+		l.GeoNameID == other.GeoNameID &&
+		l.ConnectionIP.Equal(other.ConnectionIP)
+}
+
 // NetworkAddress is the IP address with network and MAC address of a network interface
 type NetworkAddress struct {
 	NetIP netip.Prefix `gorm:"serializer:json"`
@@ -256,50 +265,88 @@ func (p *Peer) Copy() *Peer {
 	}
 }
 
-// UpdateMetaIfNew updates peer's system metadata if new information is provided
-// returns true if meta was updated, false otherwise
-func (p *Peer) UpdateMetaIfNew(ctx context.Context, meta PeerSystemMeta) (updated, versionChanged bool) {
+// UpdateMetaIfNew updates peer's system metadata and connection geo location if
+// new information is provided. newLocation is the geo location resolved from the
+// peer's current connection IP, or nil when there is nothing to apply (geo
+// disabled, no real IP, or the IP is unchanged); the caller owns the expensive
+// lookup and the same-IP guard. It returns a MetaDiff describing what changed;
+// diff.Updated() reports whether the peer needs to be persisted.
+func (p *Peer) UpdateMetaIfNew(ctx context.Context, meta PeerSystemMeta, newLocation *Location) MetaDiff {
 	if meta.isEmpty() {
-		return updated, versionChanged
+		return MetaDiff{}
 	}
 
-	versionChanged = p.Meta.WtVersion != meta.WtVersion
-
 	// Avoid overwriting UIVersion if the update was triggered sole by the CLI client
 	if meta.UIVersion == "" {
 		meta.UIVersion = p.Meta.UIVersion
 	}
 
-	oldVersion := p.Meta.WtVersion
+	effectiveLocation := p.Location
+	if newLocation != nil {
+		effectiveLocation = *newLocation
+	}
 
-	diff := metaDiff(p.Meta, meta)
-	if len(diff) != 0 {
+	diff := diffMeta(p.Meta, meta, p.Location, effectiveLocation)
+	if diff.Updated() {
 		p.Meta = meta
-		updated = true
+	}
+	p.Location = effectiveLocation
+
+	if diff.Updated() {
+		log.WithContext(ctx).Debug(diff.LogSummary())
 	}
 
-	versionInfo := ""
-	if versionChanged {
-		versionInfo = fmt.Sprintf("version changed: %s -> %s, ", oldVersion, meta.WtVersion)
-	}
-
-	if len(diff) > 0 || versionChanged {
-		log.WithContext(ctx).
-			Debugf("peer meta updated, %s%d field(s) changed: %s", versionInfo, len(diff), strings.Join(diff, ", "))
-	}
-
-	return updated, versionChanged
+	return diff
+}
+
+// MetaDiff holds a peer's full before/after state across a sync: both metas and both
+// connection locations (the location lives on Peer, not PeerSystemMeta, but posture
+// checks read it). Changed lists what moved, for logging and the persistence decision;
+// the snapshots let a posture check be replayed against old and new. Everything is derived
+// from these fields, so there are no parallel per-field flags to keep in sync.
+type MetaDiff struct {
+	OldMeta     PeerSystemMeta
+	NewMeta     PeerSystemMeta
+	OldLocation Location
+	NewLocation Location
+
+	Changed []string
+}
+
+// Updated reports whether anything changed and the peer must be persisted. diffMeta fills
+// Changed in the pass that builds the diff, so this is a length check, not a re-comparison.
+// Pointer receiver: MetaDiff embeds two metas, so copying it per call is wasteful.
+func (d *MetaDiff) Updated() bool {
+	return len(d.Changed) != 0
+}
+
+// VersionChanged reports whether the WireGuard client version changed (a client upgrade).
+func (d *MetaDiff) VersionChanged() bool {
+	return d.OldMeta.WtVersion != d.NewMeta.WtVersion
+}
+
+// HostnameChanged reports whether the peer's hostname changed.
+func (d *MetaDiff) HostnameChanged() bool {
+	return d.OldMeta.Hostname != d.NewMeta.Hostname
+}
+
+// LogSummary renders the changed fields as a single human-readable line.
+func (d *MetaDiff) LogSummary() string {
+	return fmt.Sprintf("peer meta updated, %d field(s) changed: %s",
+		len(d.Changed), strings.Join(d.Changed, ", "))
 }
 
-// metaDiff returns a human-readable list of the fields that differ between the
-// old and new meta, each formatted as `field: <old> -> <new>`. It is the single
-// source of truth for meta comparison: isEqual reports equality as an empty
-// diff, so the log line can never disagree with the change decision. Slices are
-// cloned before sorting, so callers' meta is not mutated.
 func metaDiff(oldMeta, newMeta PeerSystemMeta) []string {
-	var diff []string
+	return diffMeta(oldMeta, newMeta, Location{}, Location{}).Changed
+}
+
+// diffMeta snapshots a peer's old and new state and records a Changed entry per field that
+// moved. It is the single source of truth for the comparison: isEqual is an empty Changed
+// list, so the log line and the persistence decision can never disagree.
+func diffMeta(oldMeta, newMeta PeerSystemMeta, oldLocation, newLocation Location) MetaDiff {
+	d := MetaDiff{OldMeta: oldMeta, NewMeta: newMeta, OldLocation: oldLocation, NewLocation: newLocation}
 	add := func(field string, oldVal, newVal any) {
-		diff = append(diff, fmt.Sprintf("%s: %v -> %v", field, oldVal, newVal))
+		d.Changed = append(d.Changed, fmt.Sprintf("%s: %v -> %v", field, oldVal, newVal))
 	}
 
 	if oldMeta.Hostname != newMeta.Hostname {
@@ -353,16 +400,18 @@ func metaDiff(oldMeta, newMeta PeerSystemMeta) []string {
 	if !capabilitiesEqual(oldMeta.Capabilities, newMeta.Capabilities) {
 		add("capabilities", oldMeta.Capabilities, newMeta.Capabilities)
 	}
-
 	if !sameMultiset(oldMeta.NetworkAddresses, newMeta.NetworkAddresses) {
 		add("network_addresses", fmt.Sprintf("%v", oldMeta.NetworkAddresses), fmt.Sprintf("%v", newMeta.NetworkAddresses))
 	}
-
 	if !sameMultiset(oldMeta.Files, newMeta.Files) {
 		add("files", fmt.Sprintf("%v", oldMeta.Files), fmt.Sprintf("%v", newMeta.Files))
 	}
 
-	return diff
+	if !oldLocation.equal(newLocation) {
+		add("connection_ip", oldLocation.ConnectionIP, newLocation.ConnectionIP)
+	}
+
+	return d
 }
 
 // sameMultiset reports whether two slices contain the same elements with the

management/server/posture/affects_posture_test.go

@@ -0,0 +1,202 @@
+package posture
+
+import (
+	"context"
+	"net"
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	nbpeer "github.com/netbirdio/netbird/management/server/peer"
+)
+
+// diffFrom builds a MetaDiff from the old/new snapshots AffectsPosture replays against.
+func diffFrom(oldMeta, newMeta nbpeer.PeerSystemMeta, oldLoc, newLoc nbpeer.Location) *nbpeer.MetaDiff {
+	return &nbpeer.MetaDiff{
+		OldMeta:     oldMeta,
+		NewMeta:     newMeta,
+		OldLocation: oldLoc,
+		NewLocation: newLoc,
+	}
+}
+
+func checks(def ChecksDefinition) []*Checks {
+	return []*Checks{{Checks: def}}
+}
+
+func TestAffectsPosture_NilDiff(t *testing.T) {
+	assert.False(t, AffectsPosture(context.Background(), nil, checks(ChecksDefinition{
+		NBVersionCheck: &NBVersionCheck{MinVersion: "1.0.0"},
+	})))
+}
+
+func TestAffectsPosture_NBVersion(t *testing.T) {
+	c := checks(ChecksDefinition{NBVersionCheck: &NBVersionCheck{MinVersion: "1.2.0"}})
+
+	tests := []struct {
+		name           string
+		oldVer, newVer string
+		want           bool
+	}{
+		{"both above min, no flip", "1.3.0", "1.4.0", false},
+		{"both below min, no flip", "1.0.0", "1.1.0", false},
+		{"crosses up below->above", "1.1.0", "1.3.0", true},
+		{"crosses down above->below", "1.3.0", "1.1.0", true},
+		{"unparsable old only -> flip", "garbage", "1.3.0", true},
+		{"unparsable both -> no flip", "garbage", "junk", false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			diff := diffFrom(
+				nbpeer.PeerSystemMeta{WtVersion: tt.oldVer},
+				nbpeer.PeerSystemMeta{WtVersion: tt.newVer},
+				nbpeer.Location{}, nbpeer.Location{},
+			)
+			assert.Equal(t, tt.want, AffectsPosture(context.Background(), diff, c))
+		})
+	}
+}
+
+func TestAffectsPosture_OSVersion_KernelBumpWithinMin(t *testing.T) {
+	c := checks(ChecksDefinition{OSVersionCheck: &OSVersionCheck{
+		Linux: &MinKernelVersionCheck{MinKernelVersion: "5.0.0"},
+	}})
+
+	// Kernel moves but stays above the minimum: verdict stays pass -> not affected.
+	withinMin := diffFrom(
+		nbpeer.PeerSystemMeta{GoOS: "linux", KernelVersion: "5.10.0-arch1"},
+		nbpeer.PeerSystemMeta{GoOS: "linux", KernelVersion: "5.15.0-arch2"},
+		nbpeer.Location{}, nbpeer.Location{},
+	)
+	assert.False(t, AffectsPosture(context.Background(), withinMin, c))
+
+	// Kernel drops below the minimum: verdict flips pass -> fail -> affected.
+	crossesDown := diffFrom(
+		nbpeer.PeerSystemMeta{GoOS: "linux", KernelVersion: "5.10.0-arch1"},
+		nbpeer.PeerSystemMeta{GoOS: "linux", KernelVersion: "4.19.0-arch1"},
+		nbpeer.Location{}, nbpeer.Location{},
+	)
+	assert.True(t, AffectsPosture(context.Background(), crossesDown, c))
+}
+
+func TestAffectsPosture_OSVersion_GoOSSwitchFlipsVerdict(t *testing.T) {
+	// Only Linux is constrained. An OS outside the switch (freebsd) passes; switching to a
+	// failing linux kernel flips the verdict pass -> fail.
+	c := checks(ChecksDefinition{OSVersionCheck: &OSVersionCheck{
+		Linux: &MinKernelVersionCheck{MinKernelVersion: "6.0.0"},
+	}})
+
+	diff := diffFrom(
+		nbpeer.PeerSystemMeta{GoOS: "freebsd"},
+		nbpeer.PeerSystemMeta{GoOS: "linux", KernelVersion: "4.19.0"},
+		nbpeer.Location{}, nbpeer.Location{},
+	)
+	assert.True(t, AffectsPosture(context.Background(), diff, c))
+}
+
+func TestAffectsPosture_Process_GoOSSwitchFlipsVerdict(t *testing.T) {
+	// Process runs at a linux path. Switching GoOS to windows (no WindowsPath configured)
+	// flips the verdict.
+	c := checks(ChecksDefinition{ProcessCheck: &ProcessCheck{
+		Processes: []Process{{LinuxPath: "/usr/bin/foo"}},
+	}})
+
+	files := []nbpeer.File{{Path: "/usr/bin/foo", ProcessIsRunning: true}}
+	diff := diffFrom(
+		nbpeer.PeerSystemMeta{GoOS: "linux", Files: files},
+		nbpeer.PeerSystemMeta{GoOS: "windows", Files: files},
+		nbpeer.Location{}, nbpeer.Location{},
+	)
+	assert.True(t, AffectsPosture(context.Background(), diff, c))
+}
+
+func TestAffectsPosture_Process_UnrelatedFileChange(t *testing.T) {
+	// A tracked process stays running while an unrelated file is added: the verdict does
+	// not move, so posture is not affected.
+	c := checks(ChecksDefinition{ProcessCheck: &ProcessCheck{
+		Processes: []Process{{LinuxPath: "/usr/bin/foo"}},
+	}})
+
+	diff := diffFrom(
+		nbpeer.PeerSystemMeta{GoOS: "linux", Files: []nbpeer.File{
+			{Path: "/usr/bin/foo", ProcessIsRunning: true},
+		}},
+		nbpeer.PeerSystemMeta{GoOS: "linux", Files: []nbpeer.File{
+			{Path: "/usr/bin/foo", ProcessIsRunning: true},
+			{Path: "/usr/bin/bar", ProcessIsRunning: true},
+		}},
+		nbpeer.Location{}, nbpeer.Location{},
+	)
+	assert.False(t, AffectsPosture(context.Background(), diff, c))
+}
+
+func TestAffectsPosture_GeoLocation(t *testing.T) {
+	c := checks(ChecksDefinition{GeoLocationCheck: &GeoLocationCheck{
+		Action:    CheckActionAllow,
+		Locations: []Location{{CountryCode: "DE"}},
+	}})
+
+	// Moving within allowed countries keeps the verdict; moving out flips it.
+	stayAllowed := diffFrom(
+		nbpeer.PeerSystemMeta{}, nbpeer.PeerSystemMeta{},
+		nbpeer.Location{CountryCode: "DE", CityName: "Berlin"},
+		nbpeer.Location{CountryCode: "DE", CityName: "Munich"},
+	)
+	assert.False(t, AffectsPosture(context.Background(), stayAllowed, c))
+
+	moveOut := diffFrom(
+		nbpeer.PeerSystemMeta{}, nbpeer.PeerSystemMeta{},
+		nbpeer.Location{CountryCode: "DE"},
+		nbpeer.Location{CountryCode: "FR"},
+	)
+	assert.True(t, AffectsPosture(context.Background(), moveOut, c))
+}
+
+func TestAffectsPosture_PeerNetworkRange_ConnectionIP(t *testing.T) {
+	// The check reads the connection IP. Moving out of the allowed range flips the verdict;
+	// moving within it does not.
+	_, allowed, _ := net.ParseCIDR("10.0.0.0/8")
+	c := checks(ChecksDefinition{PeerNetworkRangeCheck: &PeerNetworkRangeCheck{
+		Action: CheckActionAllow,
+		Ranges: []netip.Prefix{netip.MustParsePrefix(allowed.String())},
+	}})
+
+	movesOutOfRange := diffFrom(
+		nbpeer.PeerSystemMeta{}, nbpeer.PeerSystemMeta{},
+		nbpeer.Location{ConnectionIP: net.ParseIP("10.1.2.3")},
+		nbpeer.Location{ConnectionIP: net.ParseIP("8.8.8.8")},
+	)
+	assert.True(t, AffectsPosture(context.Background(), movesOutOfRange, c))
+
+	staysInRange := diffFrom(
+		nbpeer.PeerSystemMeta{}, nbpeer.PeerSystemMeta{},
+		nbpeer.Location{ConnectionIP: net.ParseIP("10.1.2.3")},
+		nbpeer.Location{ConnectionIP: net.ParseIP("10.9.9.9")},
+	)
+	assert.False(t, AffectsPosture(context.Background(), staysInRange, c))
+}
+
+func TestAffectsPosture_IrrelevantFieldChange(t *testing.T) {
+	// Hostname changes but no check reads it: not affected even with checks present.
+	c := checks(ChecksDefinition{
+		NBVersionCheck:   &NBVersionCheck{MinVersion: "1.0.0"},
+		GeoLocationCheck: &GeoLocationCheck{Action: CheckActionAllow, Locations: []Location{{CountryCode: "DE"}}},
+	})
+
+	diff := diffFrom(
+		nbpeer.PeerSystemMeta{Hostname: "old", WtVersion: "1.5.0"},
+		nbpeer.PeerSystemMeta{Hostname: "new", WtVersion: "1.5.0"},
+		nbpeer.Location{CountryCode: "DE"}, nbpeer.Location{CountryCode: "DE"},
+	)
+	assert.False(t, AffectsPosture(context.Background(), diff, c))
+}
+
+func TestAffectsPosture_NoChecks(t *testing.T) {
+	diff := diffFrom(
+		nbpeer.PeerSystemMeta{WtVersion: "1.0.0"},
+		nbpeer.PeerSystemMeta{WtVersion: "2.0.0"},
+		nbpeer.Location{}, nbpeer.Location{},
+	)
+	assert.False(t, AffectsPosture(context.Background(), diff, nil))
+}

management/server/posture/checks.go

@@ -7,6 +7,8 @@ import (
 	"regexp"
 
 	"github.com/hashicorp/go-version"
+	log "github.com/sirupsen/logrus"
+
 	nbpeer "github.com/netbirdio/netbird/management/server/peer"
 	"github.com/netbirdio/netbird/shared/management/http/api"
 	"github.com/netbirdio/netbird/shared/management/status"
@@ -51,6 +53,46 @@ type Checks struct {
 	Checks ChecksDefinition `gorm:"serializer:json"`
 }
 
+// AffectsPosture reports whether the change in diff flips the verdict of any check. It
+// replays each check against the peer's old and new state and compares verdicts, so a
+// change that moves a field but stays the right side of a threshold (e.g. a kernel bump
+// still above the minimum) does not force a re-evaluation. See verdictChanged for how an
+// evaluation error counts.
+func AffectsPosture(ctx context.Context, diff *nbpeer.MetaDiff, checks []*Checks) bool {
+	if diff == nil {
+		return false
+	}
+
+	oldPeer := nbpeer.Peer{Meta: diff.OldMeta, Location: diff.OldLocation}
+	newPeer := nbpeer.Peer{Meta: diff.NewMeta, Location: diff.NewLocation}
+
+	for _, c := range checks {
+		for _, check := range c.GetChecks() {
+			if verdictChanged(ctx, check, oldPeer, newPeer) {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// verdictChanged replays check against old and new state and reports whether the verdict
+// differs. Like callers, it treats an evaluation error as deny: two errors are the same
+// verdict (no change), an error on one side only is a flip.
+func verdictChanged(ctx context.Context, check Check, oldPeer, newPeer nbpeer.Peer) bool {
+	oldPass, oldErr := check.Check(ctx, oldPeer)
+	newPass, newErr := check.Check(ctx, newPeer)
+
+	oldVerdict := oldPass && (oldErr == nil)
+	newVerdict := newPass && (newErr == nil)
+	changed := oldVerdict != newVerdict
+
+	log.WithContext(ctx).Tracef("posture check %s replay: verdict %t -> %t (changed=%t), errs: %v -> %v",
+		check.Name(), oldVerdict, newVerdict, changed, oldErr, newErr)
+
+	return changed
+}
+
 // ChecksDefinition contains definition of actual check
 type ChecksDefinition struct {
 	NBVersionCheck        *NBVersionCheck        `json:",omitempty"`

management/server/posture_checks_test.go

@@ -489,6 +489,7 @@ func TestArePostureCheckChangesAffectPeers(t *testing.T) {
 
 	policy := &types.Policy{
 		AccountID: account.Id,
+		Enabled:   true,
 		Rules: []*types.PolicyRule{
 			{
 				Enabled:      true,

management/server/store/sql_store.go

@@ -581,28 +581,6 @@ func (s *SqlStore) MarkPeerDisconnectedIfSameSession(ctx context.Context, accoun
 	return result.RowsAffected > 0, nil
 }
 
-func (s *SqlStore) SavePeerLocation(ctx context.Context, accountID string, peerWithLocation *nbpeer.Peer) error {
-	// To maintain data integrity, we create a copy of the peer's location to prevent unintended updates to other fields.
-	var peerCopy nbpeer.Peer
-	// Since the location field has been migrated to JSON serialization,
-	// updating the struct ensures the correct data format is inserted into the database.
-	peerCopy.Location = peerWithLocation.Location
-
-	result := s.db.Model(&nbpeer.Peer{}).
-		Where(accountAndIDQueryCondition, accountID, peerWithLocation.ID).
-		Updates(peerCopy)
-
-	if result.Error != nil {
-		return status.Errorf(status.Internal, "failed to save peer locations to store: %v", result.Error)
-	}
-
-	if result.RowsAffected == 0 {
-		return status.Errorf(status.NotFound, peerNotFoundFMT, peerWithLocation.ID)
-	}
-
-	return nil
-}
-
 // ApproveAccountPeers marks all peers that currently require approval in the given account as approved.
 func (s *SqlStore) ApproveAccountPeers(ctx context.Context, accountID string) (int, error) {
 	result := s.db.Model(&nbpeer.Peer{}).

management/server/store/sql_store_test.go

@@ -618,56 +618,6 @@ func TestSqlStore_SavePeerStatus(t *testing.T) {
 	assert.WithinDurationf(t, newStatus.LastSeen, actual.LastSeen.UTC(), time.Millisecond, "LastSeen should be equal")
 }
 
-func TestSqlStore_SavePeerLocation(t *testing.T) {
-	store, cleanUp, err := NewTestStoreFromSQL(context.Background(), "../testdata/store.sql", t.TempDir())
-	t.Cleanup(cleanUp)
-	assert.NoError(t, err)
-
-	account, err := store.GetAccount(context.Background(), "bf1c8084-ba50-4ce7-9439-34653001fc3b")
-	require.NoError(t, err)
-
-	peer := &nbpeer.Peer{
-		AccountID: account.Id,
-		ID:        "testpeer",
-		Location: nbpeer.Location{
-			ConnectionIP: net.ParseIP("0.0.0.0"),
-			CountryCode:  "YY",
-			CityName:     "City",
-			GeoNameID:    1,
-		},
-		CreatedAt: time.Now().UTC(),
-		Meta:      nbpeer.PeerSystemMeta{},
-	}
-	// error is expected as peer is not in store yet
-	err = store.SavePeerLocation(context.Background(), account.Id, peer)
-	assert.Error(t, err)
-
-	account.Peers[peer.ID] = peer
-	err = store.SaveAccount(context.Background(), account)
-	require.NoError(t, err)
-
-	peer.Location.ConnectionIP = net.ParseIP("35.1.1.1")
-	peer.Location.CountryCode = "DE"
-	peer.Location.CityName = "Berlin"
-	peer.Location.GeoNameID = 2950159
-
-	err = store.SavePeerLocation(context.Background(), account.Id, account.Peers[peer.ID])
-	assert.NoError(t, err)
-
-	account, err = store.GetAccount(context.Background(), account.Id)
-	require.NoError(t, err)
-
-	actual := account.Peers[peer.ID].Location
-	assert.Equal(t, peer.Location, actual)
-
-	peer.ID = "non-existing-peer"
-	err = store.SavePeerLocation(context.Background(), account.Id, peer)
-	assert.Error(t, err)
-	parsedErr, ok := status.FromError(err)
-	require.True(t, ok)
-	require.Equal(t, status.NotFound, parsedErr.Type(), "should return not found error")
-}
-
 func Test_TestGetAccountByPrivateDomain(t *testing.T) {
 	if runtime.GOOS == "windows" {
 		t.Skip("The SQLite store is not properly supported by Windows yet")

management/server/store/store.go

@@ -185,7 +185,6 @@ type Store interface {
 	// recorded by the database. Returns true when the update happened,
 	// false when a newer session has taken over.
 	MarkPeerDisconnectedIfSameSession(ctx context.Context, accountID, peerID string, sessionStartedAt int64) (bool, error)
-	SavePeerLocation(ctx context.Context, accountID string, peer *nbpeer.Peer) error
 	ApproveAccountPeers(ctx context.Context, accountID string) (int, error)
 	DeletePeer(ctx context.Context, accountID string, peerID string) error
 

management/server/store/store_mock.go

@@ -2968,20 +2968,6 @@ func (mr *MockStoreMockRecorder) SavePeer(ctx, accountID, peer interface{}) *gom
 	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "SavePeer", reflect.TypeOf((*MockStore)(nil).SavePeer), ctx, accountID, peer)
 }
 
-// SavePeerLocation mocks base method.
-func (m *MockStore) SavePeerLocation(ctx context.Context, accountID string, peer *peer.Peer) error {
-	m.ctrl.T.Helper()
-	ret := m.ctrl.Call(m, "SavePeerLocation", ctx, accountID, peer)
-	ret0, _ := ret[0].(error)
-	return ret0
-}
-
-// SavePeerLocation indicates an expected call of SavePeerLocation.
-func (mr *MockStoreMockRecorder) SavePeerLocation(ctx, accountID, peer interface{}) *gomock.Call {
-	mr.mock.ctrl.T.Helper()
-	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "SavePeerLocation", reflect.TypeOf((*MockStore)(nil).SavePeerLocation), ctx, accountID, peer)
-}
-
 // SavePeerStatus mocks base method.
 func (m *MockStore) SavePeerStatus(ctx context.Context, accountID, peerID string, status peer.PeerStatus) error {
 	m.ctrl.T.Helper()

management/server/types/peer.go

@@ -12,6 +12,9 @@ type PeerSync struct {
 	WireGuardPubKey string
 	// Meta is the system information passed by peer, must be always present
 	Meta nbpeer.PeerSystemMeta
+	// RealIP is the peer's connection IP, used to refresh its geo location.
+	// May be nil when the request has no associated connection IP.
+	RealIP net.IP
 	// UpdateAccountPeers indicate updating account peers,
 	// which occurs when the peer's metadata is updated
 	UpdateAccountPeers bool

management/server/user.go

@@ -1059,8 +1059,8 @@ func (am *DefaultAccountManager) BuildUserInfosForAccount(ctx context.Context, a
 		if err != nil {
 			return nil, err
 		}
-		log.WithContext(ctx).Debugf("Got %d users from ExternalCache for account %s", len(usersFromIntegration), accountID)
-		log.WithContext(ctx).Debugf("Got %d users from InternalCache for account %s", len(queriedUsers), accountID)
+		log.WithContext(ctx).Tracef("Got %d users from ExternalCache for account %s", len(usersFromIntegration), accountID)
+		log.WithContext(ctx).Tracef("Got %d users from InternalCache for account %s", len(queriedUsers), accountID)
 		queriedUsers = append(queriedUsers, usersFromIntegration...)
 	}
 

shared/management/status/error.go

@@ -48,6 +48,10 @@ type Type int32
 var (
 	ErrExtraSettingsNotFound = errors.New("extra settings not found")
 	ErrPeerAlreadyLoggedIn   = errors.New("peer with the same public key is already logged in")
+
+	// ErrNoAuthMethodProvided is returned when a peer login attempt carries neither a
+	// setup key nor an SSO token. Match it with errors.Is.
+	ErrNoAuthMethodProvided = Errorf(Unauthenticated, "no peer auth method provided, please use a setup key or interactive SSO login")
 )
 
 // Error is an internal error
@@ -66,6 +70,16 @@ func (e *Error) Error() string {
 	return e.Message
 }
 
+// Is reports whether target is an *Error with the same type and message,
+// enabling matching with errors.Is against sentinel errors.
+func (e *Error) Is(target error) bool {
+	var t *Error
+	if !errors.As(target, &t) {
+		return false
+	}
+	return e.ErrorType == t.ErrorType && e.Message == t.Message
+}
+
 // Errorf returns Error(ErrorType, fmt.Sprintf(format, a...)).
 func Errorf(errorType Type, format string, a ...interface{}) error {
 	return &Error{

shared/signal/client/client.go

@@ -33,7 +33,7 @@ type Client interface {
 	Receive(ctx context.Context, msgHandler func(msg *proto.Message) error) error
 	Ready() bool
 	IsHealthy() bool
-	WaitStreamConnected()
+	WaitStreamConnected(context.Context)
 	SendToStream(msg *proto.EncryptedMessage) error
 	Send(msg *proto.Message) error
 	SetOnReconnectedListener(func())

shared/signal/client/client_test.go

@@ -65,7 +65,10 @@ var _ = Describe("GrpcClient", func() {
 						return
 					}
 				}()
-				clientA.WaitStreamConnected()
+				ctxA, cancelA := context.WithTimeout(context.Background(), 5*time.Second)
+				defer cancelA()
+				clientA.WaitStreamConnected(ctxA)
+				Expect(clientA.StreamConnected()).To(BeTrue())
 
 				// connect PeerB to Signal
 				keyB, _ := wgtypes.GenerateKey()
@@ -91,7 +94,10 @@ var _ = Describe("GrpcClient", func() {
 					}
 				}()
 
-				clientB.WaitStreamConnected()
+				ctxB, cancelB := context.WithTimeout(context.Background(), 5*time.Second)
+				defer cancelB()
+				clientB.WaitStreamConnected(ctxB)
+				Expect(clientB.StreamConnected()).To(BeTrue())
 
 				// PeerA initiates ping-pong
 				err := clientA.Send(&sigProto.Message{
@@ -129,8 +135,10 @@ var _ = Describe("GrpcClient", func() {
 						return
 					}
 				}()
-				client.WaitStreamConnected()
-				Expect(client).NotTo(BeNil())
+				ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+				defer cancel()
+				client.WaitStreamConnected(ctx)
+				Expect(client.StreamConnected()).To(BeTrue())
 			})
 		})
 

shared/signal/client/grpc.go

@@ -246,15 +246,6 @@ func (c *GrpcClient) notifyStreamConnected() {
 	}
 }
 
-func (c *GrpcClient) getStreamStatusChan() <-chan struct{} {
-	c.mux.Lock()
-	defer c.mux.Unlock()
-	if c.connectedCh == nil {
-		c.connectedCh = make(chan struct{})
-	}
-	return c.connectedCh
-}
-
 func (c *GrpcClient) connect(ctx context.Context, key string) (proto.SignalExchange_ConnectStreamClient, error) {
 	c.stream = nil
 
@@ -310,14 +301,24 @@ func (c *GrpcClient) IsHealthy() bool {
 }
 
 // WaitStreamConnected waits until the client is connected to the Signal stream
-func (c *GrpcClient) WaitStreamConnected() {
-
+func (c *GrpcClient) WaitStreamConnected(ctx context.Context) {
+	// Check the status and obtain the wait channel atomically: otherwise
+	// notifyStreamConnected could flip the status and close/clear the channel
+	// between the check and the channel creation, leaving us waiting forever on
+	// a stale channel.
+	c.mux.Lock()
 	if c.status == StreamConnected {
+		c.mux.Unlock()
 		return
 	}
+	if c.connectedCh == nil {
+		c.connectedCh = make(chan struct{})
+	}
+	ch := c.connectedCh
+	c.mux.Unlock()
 
-	ch := c.getStreamStatusChan()
 	select {
+	case <-ctx.Done():
 	case <-c.ctx.Done():
 	case <-ch:
 	}

shared/signal/client/mock.go

@@ -55,7 +55,7 @@ func (sm *MockClient) Ready() bool {
 	return sm.ReadyFunc()
 }
 
-func (sm *MockClient) WaitStreamConnected() {
+func (sm *MockClient) WaitStreamConnected(context.Context) {
 	if sm.WaitStreamConnectedFunc == nil {
 		return
 	}

shared/signal/client/watchdog_test.go

@@ -65,7 +65,7 @@ func TestReceiveProbeRoundTrips(t *testing.T) {
 
 	streamReady := make(chan struct{})
 	go func() {
-		client.WaitStreamConnected()
+		client.WaitStreamConnected(ctx)
 		close(streamReady)
 	}()
 	select {