Bump actions/setup-go from 6.1.0 to 6.2.0

Bumps [actions/setup-go](https://github.com/actions/setup-go) from 6.1.0 to 6.2.0.
- [Release notes](https://github.com/actions/setup-go/releases)
- [Commits](4dc6199c7b...7a3fe6cf4c)

---
updated-dependencies:
- dependency-name: actions/setup-go
  dependency-version: 6.2.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Former-commit-id: 3d3f32d95f

.dockerignore

@@ -1,9 +1,9 @@
 .gitignore
 .dockerignore
-newt
 *.json
 README.md
 Makefile
 public/
 LICENSE
-CONTRIBUTING.md
+CONTRIBUTING.md
+bin/

.github/DISCUSSION_TEMPLATE/feature-requests.yml

@@ -0,0 +1,47 @@
+body:
+    - type: textarea
+      attributes:
+          label: Summary
+          description: A clear and concise summary of the requested feature.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Motivation
+          description: |
+              Why is this feature important?
+              Explain the problem this feature would solve or what use case it would enable.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Proposed Solution
+          description: |
+              How would you like to see this feature implemented?
+              Provide as much detail as possible about the desired behavior, configuration, or changes.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Alternatives Considered
+          description: Describe any alternative solutions or workarounds you've thought about.
+      validations:
+          required: false
+
+    - type: textarea
+      attributes:
+          label: Additional Context
+          description: Add any other context, mockups, or screenshots about the feature request here.
+      validations:
+          required: false
+
+    - type: markdown
+      attributes:
+          value: |
+              Before submitting, please:
+              - Check if there is an existing issue for this feature.
+              - Clearly explain the benefit and use case.
+              - Be as specific as possible to help contributors evaluate and implement.

.github/ISSUE_TEMPLATE/1.bug_report.yml

@@ -0,0 +1,51 @@
+name: Bug Report
+description: Create a bug report
+labels: []
+body:
+    - type: textarea
+      attributes:
+          label: Describe the Bug
+          description: A clear and concise description of what the bug is.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Environment
+          description: Please fill out the relevant details below for your environment.
+          value: |
+              - OS Type & Version: (e.g., Ubuntu 22.04)
+              - Pangolin Version:
+              - Gerbil Version:
+              - Traefik Version:
+              - Newt Version:
+              - Olm Version: (if applicable)
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: To Reproduce
+          description: |
+              Steps to reproduce the behavior, please provide a clear description of how to reproduce the issue, based on the linked minimal reproduction. Screenshots can be provided in the issue body below.
+
+              If using code blocks, make sure syntax highlighting is correct and double-check that the rendered preview is not broken.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Expected Behavior
+          description: A clear and concise description of what you expected to happen.
+      validations:
+          required: true
+
+    - type: markdown
+      attributes:
+          value: |
+              Before posting the issue go through the steps you've written down to make sure the steps provided are detailed and clear.
+
+    - type: markdown
+      attributes:
+          value: |
+              Contributors should be able to follow the steps provided in order to reproduce the bug.

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,8 @@
+blank_issues_enabled: false
+contact_links:
+    - name: Need help or have questions?
+      url: https://github.com/orgs/fosrl/discussions
+      about: Ask questions, get help, and discuss with other community members
+    - name: Request a Feature
+      url: https://github.com/orgs/fosrl/discussions/new?category=feature-requests
+      about: Feature requests should be opened as discussions so others can upvote and comment

.github/dependabot.yml

@@ -0,0 +1,30 @@
+version: 2
+updates:
+  - package-ecosystem: "gomod"
+    directory: "/"
+    schedule:
+      interval: "daily"
+    groups:
+      patch-updates:
+        update-types:
+          - "patch"
+      minor-updates:
+        update-types:
+          - "minor"
+
+  - package-ecosystem: "docker"
+    directory: "/"
+    schedule:
+      interval: "daily"
+    groups:
+      patch-updates:
+        update-types:
+          - "patch"
+      minor-updates:
+        update-types:
+          - "minor"
+          
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"

.github/workflows/cicd.yml

@@ -0,0 +1,615 @@
+name: CI/CD Pipeline
+
+permissions:
+  contents: write          # gh-release
+  packages: write          # GHCR push
+  id-token: write          # Keyless-Signatures & Attestations
+  attestations: write      # actions/attest-build-provenance
+  security-events: write   # upload-sarif
+  actions: read
+
+on:
+  push:
+    tags:
+      - "[0-9]+.[0-9]+.[0-9]+"
+      - "[0-9]+.[0-9]+.[0-9]+-rc.[0-9]+"
+
+  workflow_dispatch:
+    inputs:
+      version:
+        description: "SemVer version to release (e.g., 1.2.3, no leading 'v')"
+        required: true
+        type: string
+      publish_latest:
+        description: "Also publish the 'latest' image tag"
+        required: true
+        type: boolean
+        default: false
+      publish_minor:
+        description: "Also publish the 'major.minor' image tag (e.g., 1.2)"
+        required: true
+        type: boolean
+        default: false
+      target_branch:
+        description: "Branch to tag"
+        required: false
+        default: "main"
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event_name == 'workflow_dispatch' && github.event.inputs.version || github.ref_name }}
+  cancel-in-progress: true
+
+jobs:
+  prepare:
+    if: github.event_name == 'workflow_dispatch'
+    name: Prepare release (create tag)
+    runs-on: ubuntu-24.04
+    permissions:
+      contents: write
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6.0.1
+        with:
+          fetch-depth: 0
+
+      - name: Validate version input
+        shell: bash
+        env:
+          INPUT_VERSION: ${{ inputs.version }}
+        run: |
+          set -euo pipefail
+          if ! [[ "$INPUT_VERSION" =~ ^[0-9]+\.[0-9]+\.[0-9]+(-rc\.[0-9]+)?$ ]]; then
+            echo "Invalid version: $INPUT_VERSION (expected X.Y.Z or X.Y.Z-rc.N)" >&2
+            exit 1
+          fi
+      - name: Create and push tag
+        shell: bash
+        env:
+          TARGET_BRANCH: ${{ inputs.target_branch }}
+          VERSION: ${{ inputs.version }}
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          set -euo pipefail
+          git config user.name  "github-actions[bot]"
+          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
+          git fetch --prune origin
+          git checkout "$TARGET_BRANCH"
+          git pull --ff-only origin "$TARGET_BRANCH"
+          if git rev-parse -q --verify "refs/tags/$VERSION" >/dev/null; then
+            echo "Tag $VERSION already exists" >&2
+            exit 1
+          fi
+          git tag -a "$VERSION" -m "Release $VERSION"
+          git push origin "refs/tags/$VERSION"
+  release:
+    if: ${{ github.event_name == 'workflow_dispatch' || (github.event_name == 'push' && github.actor != 'github-actions[bot]') }}
+    name: Build and Release
+    runs-on: ubuntu-24.04
+    timeout-minutes: 120
+    env:
+      DOCKERHUB_IMAGE: docker.io/fosrl/${{ github.event.repository.name }}
+      GHCR_IMAGE: ghcr.io/${{ github.repository_owner }}/${{ github.event.repository.name }}
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6.0.1
+        with:
+          fetch-depth: 0
+
+      - name: Capture created timestamp
+        run: echo "IMAGE_CREATED=$(date -u +%Y-%m-%dT%H:%M:%SZ)" >> $GITHUB_ENV
+        shell: bash
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@c7c53464625b32c7a7e944ae62b3e17d2b600130 # v3.7.0
+
+      - name: Set up 1.2.0 Buildx
+        uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3.12.0
+
+      - name: Log in to Docker Hub
+        uses: docker/login-action@5e57cd118135c172c3672efd75eb46360885c0ef # v3.6.0
+        with:
+          registry: docker.io
+          username: ${{ secrets.DOCKER_HUB_USERNAME }}
+          password: ${{ secrets.DOCKER_HUB_ACCESS_TOKEN }}
+
+      - name: Log in to GHCR
+        uses: docker/login-action@5e57cd118135c172c3672efd75eb46360885c0ef # v3.6.0
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Normalize image names to lowercase
+        run: |
+          set -euo pipefail
+          echo "GHCR_IMAGE=${GHCR_IMAGE,,}" >> "$GITHUB_ENV"
+          echo "DOCKERHUB_IMAGE=${DOCKERHUB_IMAGE,,}" >> "$GITHUB_ENV"
+        shell: bash
+
+      - name: Extract tag name
+        env:
+          EVENT_NAME: ${{ github.event_name }}
+          INPUT_VERSION: ${{ inputs.version }}
+        run: |
+          if [ "$EVENT_NAME" = "workflow_dispatch" ]; then
+            echo "TAG=${INPUT_VERSION}" >> $GITHUB_ENV
+          else
+            echo "TAG=${{ github.ref_name }}" >> $GITHUB_ENV
+          fi
+        shell: bash
+
+      - name: Validate pushed tag format (no leading 'v')
+        if: ${{ github.event_name == 'push' }}
+        shell: bash
+        env:
+          TAG_GOT: ${{ env.TAG }}
+        run: |
+          set -euo pipefail
+          if [[ "$TAG_GOT" =~ ^[0-9]+\.[0-9]+\.[0-9]+(-rc\.[0-9]+)?$ ]]; then
+            echo "Tag OK: $TAG_GOT"
+            exit 0
+          fi
+          echo "ERROR: Tag '$TAG_GOT' is not allowed. Use 'X.Y.Z' or 'X.Y.Z-rc.N' (no leading 'v')." >&2
+          exit 1
+      - name: Wait for tag to be visible (dispatch only)
+        if: ${{ github.event_name == 'workflow_dispatch' }}
+        run: |
+          set -euo pipefail
+          for i in {1..90}; do
+            if git ls-remote --tags origin "refs/tags/${TAG}" | grep -qE "refs/tags/${TAG}$"; then
+              echo "Tag ${TAG} is visible on origin"; exit 0
+            fi
+            echo "Tag not yet visible, retrying... ($i/90)"
+            sleep 2
+          done
+          echo "Tag ${TAG} not visible after waiting"; exit 1
+        shell: bash
+
+      - name: Update version in main.go
+        run: |
+            TAG=${{ env.TAG }}
+            if [ -f main.go ]; then
+            sed -i 's/version_replaceme/'"$TAG"'/' main.go
+            echo "Updated main.go with version $TAG"
+            else
+            echo "main.go not found"
+            fi
+
+      - name: Ensure repository is at the tagged commit (dispatch only)
+        if: ${{ github.event_name == 'workflow_dispatch' }}
+        run: |
+          set -euo pipefail
+          git fetch --tags --force
+          git checkout "refs/tags/${TAG}"
+          echo "Checked out $(git rev-parse --short HEAD) for tag ${TAG}"
+        shell: bash
+
+      - name: Detect release candidate (rc)
+        run: |
+          set -euo pipefail
+          if [[ "${TAG}" =~ ^[0-9]+\.[0-9]+\.[0-9]+-rc\.[0-9]+$ ]]; then
+            echo "IS_RC=true" >> $GITHUB_ENV
+          else
+            echo "IS_RC=false" >> $GITHUB_ENV
+          fi
+        shell: bash
+
+      - name: Install Go
+        uses: actions/setup-go@7a3fe6cf4cb3a834922a1244abfce67bcef6a0c5 # v6.2.0
+        with:
+          go-version-file: go.mod
+
+      - name: Resolve publish-latest flag
+        env:
+          EVENT_NAME: ${{ github.event_name }}
+          PL_INPUT: ${{ inputs.publish_latest }}
+          PL_VAR: ${{ vars.PUBLISH_LATEST }}
+        run: |
+          set -euo pipefail
+          val="false"
+          if [ "$EVENT_NAME" = "workflow_dispatch" ]; then
+            if [ "${PL_INPUT}" = "true" ]; then val="true"; fi
+          else
+            if [ "${PL_VAR}" = "true" ]; then val="true"; fi
+          fi
+          echo "PUBLISH_LATEST=$val" >> $GITHUB_ENV
+        shell: bash
+
+      - name: Resolve publish-minor flag
+        env:
+          EVENT_NAME: ${{ github.event_name }}
+          PM_INPUT: ${{ inputs.publish_minor }}
+          PM_VAR: ${{ vars.PUBLISH_MINOR }}
+        run: |
+          set -euo pipefail
+          val="false"
+          if [ "$EVENT_NAME" = "workflow_dispatch" ]; then
+            if [ "${PM_INPUT}" = "true" ]; then val="true"; fi
+          else
+            if [ "${PM_VAR}" = "true" ]; then val="true"; fi
+          fi
+          echo "PUBLISH_MINOR=$val" >> $GITHUB_ENV
+        shell: bash
+
+      - name: Cache Go modules
+        if: ${{ hashFiles('**/go.sum') != '' }}
+        uses: actions/cache@8b402f58fbc84540c8b491a91e594a4576fec3d7 # v5.0.2
+        with:
+          path: |
+            ~/.cache/go-build
+            ~/go/pkg/mod
+          key: ${{ runner.os }}-go-${{ hashFiles('**/go.sum') }}
+          restore-keys: |
+            ${{ runner.os }}-go-
+      - name: Go vet & test
+        if: ${{ hashFiles('**/go.mod') != '' }}
+        run: |
+          go version
+          go vet ./...
+          go test ./... -race -covermode=atomic
+        shell: bash
+
+      - name: Resolve license fallback
+        run: echo "IMAGE_LICENSE=${{ github.event.repository.license.spdx_id || 'NOASSERTION' }}" >> $GITHUB_ENV
+        shell: bash
+
+      - name: Resolve registries list (GHCR always, Docker Hub only if creds)
+        shell: bash
+        run: |
+          set -euo pipefail
+          images="${GHCR_IMAGE}"
+          if [ -n "${{ secrets.DOCKER_HUB_ACCESS_TOKEN }}" ] && [ -n "${{ secrets.DOCKER_HUB_USERNAME }}" ]; then
+            images="${images}\n${DOCKERHUB_IMAGE}"
+          fi
+          {
+            echo 'IMAGE_LIST<<EOF'
+            echo -e "$images"
+            echo 'EOF'
+          } >> "$GITHUB_ENV"
+      - name: Docker meta
+        id: meta
+        uses: docker/metadata-action@c299e40c65443455700f0fdfc63efafe5b349051 # v5.10.0
+        with:
+          images: ${{ env.IMAGE_LIST }}
+          tags: |
+            type=semver,pattern={{version}},value=${{ env.TAG }}
+            type=semver,pattern={{major}}.{{minor}},value=${{ env.TAG }},enable=${{ env.PUBLISH_MINOR == 'true' && env.IS_RC != 'true' }}
+            type=raw,value=latest,enable=${{ env.IS_RC != 'true' }}
+          flavor: |
+            latest=false
+          labels: |
+            org.opencontainers.image.title=${{ github.event.repository.name }}
+            org.opencontainers.image.version=${{ env.TAG }}
+            org.opencontainers.image.revision=${{ github.sha }}
+            org.opencontainers.image.source=${{ github.event.repository.html_url }}
+            org.opencontainers.image.url=${{ github.event.repository.html_url }}
+            org.opencontainers.image.documentation=${{ github.event.repository.html_url }}
+            org.opencontainers.image.description=${{ github.event.repository.description }}
+            org.opencontainers.image.licenses=${{ env.IMAGE_LICENSE }}
+            org.opencontainers.image.created=${{ env.IMAGE_CREATED }}
+            org.opencontainers.image.ref.name=${{ env.TAG }}
+            org.opencontainers.image.authors=${{ github.repository_owner }}
+      - name: Echo build config (non-secret)
+        shell: bash
+        env:
+          IMAGE_TITLE:        ${{ github.event.repository.name }}
+          IMAGE_VERSION:      ${{ env.TAG }}
+          IMAGE_REVISION:     ${{ github.sha }}
+          IMAGE_SOURCE_URL:   ${{ github.event.repository.html_url }}
+          IMAGE_URL:          ${{ github.event.repository.html_url }}
+          IMAGE_DESCRIPTION:  ${{ github.event.repository.description }}
+          IMAGE_LICENSE:      ${{ env.IMAGE_LICENSE }}
+          DOCKERHUB_IMAGE:    ${{ env.DOCKERHUB_IMAGE }}
+          GHCR_IMAGE:         ${{ env.GHCR_IMAGE }}
+          DOCKER_HUB_USER:    ${{ secrets.DOCKER_HUB_USERNAME }}
+          REPO:               ${{ github.repository }}
+          OWNER:              ${{ github.repository_owner }}
+          WORKFLOW_REF:       ${{ github.workflow_ref }}
+          REF:                ${{ github.ref }}
+          REF_NAME:           ${{ github.ref_name }}
+          RUN_URL:            https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}
+        run: |
+          set -euo pipefail
+          echo "=== OCI Label Values ==="
+          echo "org.opencontainers.image.title=${IMAGE_TITLE}"
+          echo "org.opencontainers.image.version=${IMAGE_VERSION}"
+          echo "org.opencontainers.image.revision=${IMAGE_REVISION}"
+          echo "org.opencontainers.image.source=${IMAGE_SOURCE_URL}"
+          echo "org.opencontainers.image.url=${IMAGE_URL}"
+          echo "org.opencontainers.image.description=${IMAGE_DESCRIPTION}"
+          echo "org.opencontainers.image.licenses=${IMAGE_LICENSE}"
+          echo
+          echo "=== Images ==="
+          echo "DOCKERHUB_IMAGE=${DOCKERHUB_IMAGE}"
+          echo "GHCR_IMAGE=${GHCR_IMAGE}"
+          echo "DOCKER_HUB_USERNAME=${DOCKER_HUB_USER}"
+          echo
+          echo "=== GitHub Kontext ==="
+          echo "repository=${REPO}"
+          echo "owner=${OWNER}"
+          echo "workflow_ref=${WORKFLOW_REF}"
+          echo "ref=${REF}"
+          echo "ref_name=${REF_NAME}"
+          echo "run_url=${RUN_URL}"
+          echo
+          echo "=== docker/metadata-action outputs (Tags/Labels), raw ==="
+          echo "::group::tags"
+          echo "${{ steps.meta.outputs.tags }}"
+          echo "::endgroup::"
+          echo "::group::labels"
+          echo "${{ steps.meta.outputs.labels }}"
+          echo "::endgroup::"
+      - name: Build and push (Docker Hub + GHCR)
+        id: build
+        uses: docker/build-push-action@263435318d21b8e681c14492fe198d362a7d2c83 # v6.18.0
+        with:
+          context: .
+          push: true
+          platforms: linux/amd64,linux/arm64,linux/arm/v7
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          cache-from: type=gha,scope=${{ github.repository }}
+          cache-to: type=gha,mode=max,scope=${{ github.repository }}
+          provenance: mode=max
+          sbom: true
+
+      - name: Compute image digest refs
+        run: |
+          echo "DIGEST=${{ steps.build.outputs.digest }}" >> $GITHUB_ENV
+          echo "GHCR_REF=$GHCR_IMAGE@${{ steps.build.outputs.digest }}" >> $GITHUB_ENV
+          echo "DH_REF=$DOCKERHUB_IMAGE@${{ steps.build.outputs.digest }}" >> $GITHUB_ENV
+          echo "Built digest: ${{ steps.build.outputs.digest }}"
+        shell: bash
+
+      - name: Attest build provenance (GHCR)
+        id: attest-ghcr
+        uses: actions/attest-build-provenance@977bb373ede98d70efdf65b84cb5f73e068dcc2a # v3.0.0
+        with:
+          subject-name: ${{ env.GHCR_IMAGE }}
+          subject-digest: ${{ steps.build.outputs.digest }}
+          push-to-registry: true
+          show-summary: true
+
+      - name: Attest build provenance (Docker Hub)
+        continue-on-error: true
+        id: attest-dh
+        uses: actions/attest-build-provenance@977bb373ede98d70efdf65b84cb5f73e068dcc2a # v3.0.0
+        with:
+          subject-name: index.docker.io/fosrl/${{ github.event.repository.name }}
+          subject-digest: ${{ steps.build.outputs.digest }}
+          push-to-registry: true
+          show-summary: true
+
+      - name: Install cosign
+        uses: sigstore/cosign-installer@faadad0cce49287aee09b3a48701e75088a2c6ad # v4.0.0
+        with:
+          cosign-release: 'v3.0.2'
+
+      - name: Sanity check cosign private key
+        env:
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+        run: |
+          set -euo pipefail
+          cosign public-key --key env://COSIGN_PRIVATE_KEY >/dev/null
+        shell: bash
+
+      - name: Sign GHCR image (digest) with key (recursive)
+        env:
+          COSIGN_YES: "true"
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+        run: |
+          set -euo pipefail
+          echo "Signing ${GHCR_REF} (digest) recursively with provided key"
+          cosign sign --key env://COSIGN_PRIVATE_KEY --recursive "${GHCR_REF}"
+          echo "Waiting 30 seconds for signatures to propagate..."
+        shell: bash
+
+      - name: Generate SBOM (SPDX JSON)
+        uses: aquasecurity/trivy-action@b6643a29fecd7f34b3597bc6acb0a98b03d33ff8 # v0.33.1
+        with:
+          image-ref: ${{ env.GHCR_IMAGE }}@${{ steps.build.outputs.digest }}
+          format: spdx-json
+          output: sbom.spdx.json
+
+      - name: Validate SBOM JSON
+        run: jq -e . sbom.spdx.json >/dev/null
+        shell: bash
+
+      - name: Minify SBOM JSON (optional hardening)
+        run: jq -c . sbom.spdx.json > sbom.min.json && mv sbom.min.json sbom.spdx.json
+        shell: bash
+
+      - name: Create SBOM attestation (GHCR, private key)
+        env:
+          COSIGN_YES: "true"
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+        run: |
+          set -euo pipefail
+          cosign attest \
+            --key env://COSIGN_PRIVATE_KEY \
+            --type spdxjson \
+            --predicate sbom.spdx.json \
+            "${GHCR_REF}"
+        shell: bash
+
+      - name: Create SBOM attestation (Docker Hub, private key)
+        continue-on-error: true
+        env:
+          COSIGN_YES: "true"
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+          COSIGN_DOCKER_MEDIA_TYPES: "1"
+        run: |
+          set -euo pipefail
+          cosign attest \
+            --key env://COSIGN_PRIVATE_KEY \
+            --type spdxjson \
+            --predicate sbom.spdx.json \
+            "${DH_REF}"
+        shell: bash
+
+      - name: Keyless sign & verify GHCR digest (OIDC)
+        env:
+          COSIGN_YES: "true"
+          WORKFLOW_REF: ${{ github.workflow_ref }}  # owner/repo/.github/workflows/<file>@refs/tags/<tag>
+          ISSUER: https://token.actions.githubusercontent.com
+        run: |
+          set -euo pipefail
+          echo "Keyless signing ${GHCR_REF}"
+          cosign sign --rekor-url https://rekor.sigstore.dev --recursive "${GHCR_REF}"
+          echo "Verify keyless (OIDC) signature policy on ${GHCR_REF}"
+          cosign verify \
+            --certificate-oidc-issuer "${ISSUER}" \
+            --certificate-identity "https://github.com/${WORKFLOW_REF}" \
+            "${GHCR_REF}" -o text
+        shell: bash
+
+      - name: Sign Docker Hub image (digest) with key (recursive)
+        continue-on-error: true
+        env:
+          COSIGN_YES: "true"
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+          COSIGN_DOCKER_MEDIA_TYPES: "1"
+        run: |
+          set -euo pipefail
+          echo "Signing ${DH_REF} (digest) recursively with provided key (Docker media types fallback)"
+          cosign sign --key env://COSIGN_PRIVATE_KEY --recursive "${DH_REF}"
+        shell: bash
+
+      - name: Keyless sign & verify Docker Hub digest (OIDC)
+        continue-on-error: true
+        env:
+          COSIGN_YES: "true"
+          ISSUER: https://token.actions.githubusercontent.com
+          COSIGN_DOCKER_MEDIA_TYPES: "1"
+        run: |
+          set -euo pipefail
+          echo "Keyless signing ${DH_REF} (force public-good Rekor)"
+          cosign sign --rekor-url https://rekor.sigstore.dev --recursive "${DH_REF}"
+          echo "Keyless verify via Rekor (strict identity)"
+          if ! cosign verify \
+            --rekor-url https://rekor.sigstore.dev \
+            --certificate-oidc-issuer "${ISSUER}" \
+            --certificate-identity "https://github.com/${{ github.workflow_ref }}" \
+            "${DH_REF}" -o text; then
+            echo "Rekor verify failed — retry offline bundle verify (no Rekor)"
+            if ! cosign verify \
+              --offline \
+              --certificate-oidc-issuer "${ISSUER}" \
+              --certificate-identity "https://github.com/${{ github.workflow_ref }}" \
+              "${DH_REF}" -o text; then
+              echo "Offline bundle verify failed — ignore tlog (TEMP for debugging)"
+              cosign verify \
+                --insecure-ignore-tlog=true \
+                --certificate-oidc-issuer "${ISSUER}" \
+                --certificate-identity "https://github.com/${{ github.workflow_ref }}" \
+                "${DH_REF}" -o text || true
+            fi
+          fi
+      - name: Verify signature (public key) GHCR digest + tag
+        env:
+          COSIGN_PUBLIC_KEY: ${{ secrets.COSIGN_PUBLIC_KEY }}
+        run: |
+          set -euo pipefail
+          TAG_VAR="${TAG}"
+          echo "Verifying (digest) ${GHCR_REF}"
+          cosign verify --key env://COSIGN_PUBLIC_KEY "$GHCR_REF" -o text
+          echo "Verifying (tag) $GHCR_IMAGE:$TAG_VAR"
+          cosign verify --key env://COSIGN_PUBLIC_KEY "$GHCR_IMAGE:$TAG_VAR" -o text
+        shell: bash
+
+      - name: Verify SBOM attestation (GHCR)
+        env:
+          COSIGN_PUBLIC_KEY: ${{ secrets.COSIGN_PUBLIC_KEY }}
+        run: cosign verify-attestation --key env://COSIGN_PUBLIC_KEY --type spdxjson "$GHCR_REF" -o text
+        shell: bash
+
+      - name: Verify SLSA provenance (GHCR)
+        env:
+          ISSUER: https://token.actions.githubusercontent.com
+          WFREF: ${{ github.workflow_ref }}
+        run: |
+          set -euo pipefail
+          # (optional) show which predicate types are present to aid debugging
+          cosign download attestation "$GHCR_REF" \
+            | jq -r '.payload | @base64d | fromjson | .predicateType' | sort -u || true
+          # Verify the SLSA v1 provenance attestation (predicate URL)
+          cosign verify-attestation \
+            --type 'https://slsa.dev/provenance/v1' \
+            --certificate-oidc-issuer "$ISSUER" \
+            --certificate-identity "https://github.com/${WFREF}" \
+            --rekor-url https://rekor.sigstore.dev \
+            "$GHCR_REF" -o text
+        shell: bash
+
+      - name: Verify signature (public key) Docker Hub digest
+        continue-on-error: true
+        env:
+          COSIGN_PUBLIC_KEY: ${{ secrets.COSIGN_PUBLIC_KEY }}
+          COSIGN_DOCKER_MEDIA_TYPES: "1"
+        run: |
+          set -euo pipefail
+          echo "Verifying (digest) ${DH_REF} with Docker media types"
+          cosign verify --key env://COSIGN_PUBLIC_KEY "${DH_REF}" -o text
+        shell: bash
+
+      - name: Verify signature (public key) Docker Hub tag
+        continue-on-error: true
+        env:
+          COSIGN_PUBLIC_KEY: ${{ secrets.COSIGN_PUBLIC_KEY }}
+          COSIGN_DOCKER_MEDIA_TYPES: "1"
+        run: |
+          set -euo pipefail
+          echo "Verifying (tag) $DOCKERHUB_IMAGE:$TAG with Docker media types"
+          cosign verify --key env://COSIGN_PUBLIC_KEY "$DOCKERHUB_IMAGE:$TAG" -o text
+        shell: bash
+
+      # - name: Trivy scan (GHCR image)
+      #   id: trivy
+      #   uses: aquasecurity/trivy-action@b6643a29fecd7f34b3597bc6acb0a98b03d33ff8 # v0.33.1
+      #   with:
+      #     image-ref: ${{ env.GHCR_IMAGE }}@${{ steps.build.outputs.digest }}
+      #     format: sarif
+      #     output: trivy-ghcr.sarif
+      #     ignore-unfixed: true
+      #     vuln-type: os,library
+      #     severity: CRITICAL,HIGH
+      #     exit-code: ${{ (vars.TRIVY_FAIL || '0') }}
+
+      # - name: Upload SARIF
+      #   if: ${{ always() && hashFiles('trivy-ghcr.sarif') != '' }}
+      #   uses: github/codeql-action/upload-sarif@fdbfb4d2750291e159f0156def62b853c2798ca2 # v4.31.5
+      #   with:
+      #     sarif_file: trivy-ghcr.sarif
+      #     category: Image Vulnerability Scan
+
+      - name: Build binaries
+        env:
+          CGO_ENABLED: "0"
+          GOFLAGS: "-trimpath"
+        run: |
+          set -euo pipefail
+          TAG_VAR="${TAG}"
+          make go-build-release tag=$TAG_VAR
+        shell: bash
+
+      - name: Create GitHub Release
+        uses: softprops/action-gh-release@a06a81a03ee405af7f2048a818ed3f03bbf83c7b # v2.5.0
+        with:
+          tag_name: ${{ env.TAG }}
+          generate_release_notes: true
+          prerelease: ${{ env.IS_RC == 'true' }}
+          files: |
+            bin/*
+          fail_on_unmatched_files: true
+          draft: true
+          body: |
+            ## Container Images
+            - GHCR: `${{ env.GHCR_REF }}`
+            - Docker Hub: `${{ env.DH_REF || 'N/A' }}`
+            **Digest:** `${{ steps.build.outputs.digest }}`

.github/workflows/mirror.yaml

@@ -0,0 +1,132 @@
+name: Mirror & Sign (Docker Hub to GHCR)
+
+on:
+  workflow_dispatch: {}
+
+permissions:
+  contents: read
+  packages: write
+  id-token: write   # for keyless OIDC
+
+env:
+  SOURCE_IMAGE: docker.io/fosrl/olm
+  DEST_IMAGE: ghcr.io/${{ github.repository_owner }}/${{ github.event.repository.name }}
+
+jobs:
+  mirror-and-dual-sign:
+    runs-on: amd64-runner
+    steps:
+      - name: Install skopeo + jq
+        run: |
+          sudo apt-get update -y
+          sudo apt-get install -y skopeo jq
+          skopeo --version
+
+      - name: Install cosign
+        uses: sigstore/cosign-installer@faadad0cce49287aee09b3a48701e75088a2c6ad # v4.0.0
+
+      - name: Input check
+        run: |
+          test -n "${SOURCE_IMAGE}" || (echo "SOURCE_IMAGE is empty" && exit 1)
+          echo "Source : ${SOURCE_IMAGE}"
+          echo "Target : ${DEST_IMAGE}"
+
+      # Auth for skopeo (containers-auth)
+      - name: Skopeo login to GHCR
+        run: |
+          skopeo login ghcr.io -u "${{ github.actor }}" -p "${{ secrets.GITHUB_TOKEN }}"
+
+      # Auth for cosign (docker-config)
+      - name: Docker login to GHCR (for cosign)
+        run: |
+          echo "${{ secrets.GITHUB_TOKEN }}" | docker login ghcr.io -u "${{ github.actor }}" --password-stdin
+
+      - name: List source tags
+        run: |
+          set -euo pipefail
+          skopeo list-tags --retry-times 3 docker://"${SOURCE_IMAGE}" \
+            | jq -r '.Tags[]' | sort -u > src-tags.txt
+          echo "Found source tags: $(wc -l < src-tags.txt)"
+          head -n 20 src-tags.txt || true
+
+      - name: List destination tags (skip existing)
+        run: |
+          set -euo pipefail
+          if skopeo list-tags --retry-times 3 docker://"${DEST_IMAGE}" >/tmp/dst.json 2>/dev/null; then
+            jq -r '.Tags[]' /tmp/dst.json | sort -u > dst-tags.txt
+          else
+            : > dst-tags.txt
+          fi
+          echo "Existing destination tags: $(wc -l < dst-tags.txt)"
+
+      - name: Mirror, dual-sign, and verify
+        env:
+          # keyless
+          COSIGN_YES: "true"
+          # key-based
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+          # verify
+          COSIGN_PUBLIC_KEY:  ${{ secrets.COSIGN_PUBLIC_KEY }}
+        run: |
+          set -euo pipefail
+          copied=0; skipped=0; v_ok=0; errs=0
+
+          issuer="https://token.actions.githubusercontent.com"
+          id_regex="^https://github.com/${{ github.repository }}/.+"
+
+          while read -r tag; do
+            [ -z "$tag" ] && continue
+
+            if grep -Fxq "$tag" dst-tags.txt; then
+              echo "::notice ::Skip (exists) ${DEST_IMAGE}:${tag}"
+              skipped=$((skipped+1))
+              continue
+            fi
+
+            echo "==> Copy ${SOURCE_IMAGE}:${tag} → ${DEST_IMAGE}:${tag}"
+            if ! skopeo copy --all --retry-times 3 \
+                 docker://"${SOURCE_IMAGE}:${tag}" docker://"${DEST_IMAGE}:${tag}"; then
+              echo "::warning title=Copy failed::${SOURCE_IMAGE}:${tag}"
+              errs=$((errs+1)); continue
+            fi
+            copied=$((copied+1))
+
+            digest="$(skopeo inspect --retry-times 3 docker://"${DEST_IMAGE}:${tag}" | jq -r '.Digest')"
+            ref="${DEST_IMAGE}@${digest}"
+
+            echo "==> cosign sign (keyless) --recursive ${ref}"
+            if ! cosign sign --recursive "${ref}"; then
+              echo "::warning title=Keyless sign failed::${ref}"
+              errs=$((errs+1))
+            fi
+
+            echo "==> cosign sign (key) --recursive ${ref}"
+            if ! cosign sign --key env://COSIGN_PRIVATE_KEY --recursive "${ref}"; then
+              echo "::warning title=Key sign failed::${ref}"
+              errs=$((errs+1))
+            fi
+
+            echo "==> cosign verify (public key) ${ref}"
+            if ! cosign verify --key env://COSIGN_PUBLIC_KEY "${ref}" -o text; then
+              echo "::warning title=Verify(pubkey) failed::${ref}"
+              errs=$((errs+1))
+            fi
+
+            echo "==> cosign verify (keyless policy) ${ref}"
+            if ! cosign verify \
+                 --certificate-oidc-issuer "${issuer}" \
+                 --certificate-identity-regexp "${id_regex}" \
+                 "${ref}" -o text; then
+              echo "::warning title=Verify(keyless) failed::${ref}"
+              errs=$((errs+1))
+            else
+              v_ok=$((v_ok+1))
+            fi
+          done < src-tags.txt
+
+          echo "---- Summary ----"
+          echo "Copied        : $copied"
+          echo "Skipped       : $skipped"
+          echo "Verified OK   : $v_ok"
+          echo "Errors        : $errs"

.github/workflows/stale-bot.yml

@@ -0,0 +1,37 @@
+name: Mark and Close Stale Issues
+
+on:
+  schedule:
+    - cron: '0 0 * * *'
+  workflow_dispatch: # Allow manual trigger
+
+permissions:
+  contents: write # only for delete-branch option
+  issues: write
+  pull-requests: write
+
+jobs:
+  stale:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/stale@997185467fa4f803885201cee163a9f38240193d # v10.1.1
+        with:
+          days-before-stale: 14
+          days-before-close: 14
+          stale-issue-message: 'This issue has been automatically marked as stale due to 14 days of inactivity. It will be closed in 14 days if no further activity occurs.'
+          close-issue-message: 'This issue has been automatically closed due to inactivity. If you believe this is still relevant, please open a new issue with up-to-date information.'
+          stale-issue-label: 'stale'
+          
+          exempt-issue-labels: 'needs investigating, networking, new feature, reverse proxy, bug, api, authentication, documentation, enhancement, help wanted, good first issue, question'
+          
+          exempt-all-issue-assignees: true
+          
+          only-labels: ''
+          exempt-pr-labels: ''
+          days-before-pr-stale: -1
+          days-before-pr-close: -1
+          
+          operations-per-run: 100
+          remove-stale-when-updated: true
+          delete-branch: false
+          enable-statistics: true

.github/workflows/test.yml

@@ -0,0 +1,42 @@
+name: Run Tests
+
+permissions:
+  contents: read
+
+on:
+  pull_request:
+    branches:
+      - main
+      - dev
+
+jobs:
+  build-go:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6.0.1
+
+      - name: Set up Go
+        uses: actions/setup-go@7a3fe6cf4cb3a834922a1244abfce67bcef6a0c5 # v6.2.0
+        with:
+          go-version: 1.25
+
+      - name: Build binaries
+        run: make go-build-release
+
+  build-docker:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6.0.1
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@c7c53464625b32c7a7e944ae62b3e17d2b600130 # v3.7.0
+
+      - name: Set up 1.2.0 Buildx
+        uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3.12.0
+
+      - name: Build Docker image
+        run: make docker-build-dev

.gitignore

@@ -1 +1,2 @@
-newt
+.DS_Store
+bin/

.go-version

@@ -0,0 +1 @@
+1.25

API.md

@@ -0,0 +1,306 @@
+## API
+
+Olm can be controlled with an embedded API server when using `--enable-api`. This allows you to start it as a daemon and trigger it with the following endpoints. The API can listen on either a TCP address or a Unix socket/Windows named pipe.
+
+### Socket vs TCP
+
+When `--enable-api` is used, Olm can listen on a TCP address when configured via `--http-addr` (like `:9452`). Alternatively, Olm can listen on a Unix socket (Linux/macOS) or Windows named pipe for local-only communication with better security when using `--socket-path` (like `/var/run/olm.sock`).
+
+**Unix Socket (Linux/macOS):**
+- Socket path example: `/var/run/olm/olm.sock`
+- The directory is created automatically if it doesn't exist
+- Socket permissions are set to `0666` to allow access
+- Existing socket files are automatically removed on startup
+- Socket file is cleaned up when Olm stops
+
+**Windows Named Pipe:**
+- Pipe path example: `\\.\pipe\olm`
+- If the path doesn't start with `\`, it's automatically prefixed with `\\.\pipe\`
+- Security descriptor grants full access to Everyone and the current owner
+- Named pipes are automatically cleaned up by Windows
+
+**Connecting to the Socket:**
+
+```bash
+# Linux/macOS - using curl with Unix socket
+curl --unix-socket /var/run/olm/olm.sock http://localhost/status
+
+---
+
+### POST /connect
+Initiates a new connection request to a Pangolin server.
+
+**Request Body:**
+```json
+{
+  "id": "string",
+  "secret": "string",
+  "endpoint": "string",
+  "userToken": "string",
+  "mtu": 1280,
+  "dns": "8.8.8.8",
+  "dnsProxyIP": "string",
+  "upstreamDNS": ["8.8.8.8:53", "1.1.1.1:53"],
+  "interfaceName": "olm",
+  "holepunch": false,
+  "tlsClientCert": "string",
+  "pingInterval": "3s",
+  "pingTimeout": "5s",
+  "orgId": "string"
+}
+```
+
+**Required Fields:**
+- `id`: Olm ID generated by Pangolin
+- `secret`: Authentication secret for the Olm ID
+- `endpoint`: Target Pangolin endpoint URL
+
+**Optional Fields:**
+- `userToken`: User authentication token
+- `mtu`: MTU for the internal WireGuard interface (default: 1280)
+- `dns`: DNS server to use for resolving the endpoint
+- `dnsProxyIP`: DNS proxy IP address
+- `upstreamDNS`: Array of upstream DNS servers
+- `interfaceName`: Name of the WireGuard interface (default: olm)
+- `holepunch`: Enable NAT hole punching (default: false)
+- `tlsClientCert`: TLS client certificate
+- `pingInterval`: Interval for pinging the server (default: 3s)
+- `pingTimeout`: Timeout for each ping (default: 5s)
+- `orgId`: Organization ID to connect to
+
+**Response:**
+- **Status Code:** `202 Accepted`
+- **Content-Type:** `application/json`
+
+```json
+{
+  "status": "connection request accepted"
+}
+```
+
+**Error Responses:**
+- `405 Method Not Allowed` - Non-POST requests
+- `400 Bad Request` - Invalid JSON or missing required fields
+- `409 Conflict` - Already connected to a server (disconnect first)
+
+---
+
+### GET /status
+Returns the current connection status, registration state, and peer information.
+
+**Response:**
+- **Status Code:** `200 OK`
+- **Content-Type:** `application/json`
+
+```json
+{
+  "connected": true,
+  "registered": true,
+  "terminated": false,
+  "version": "1.0.0",
+  "agent": "olm",
+  "orgId": "org_123",
+  "peers": {
+    "10": {
+      "siteId": 10,
+      "name": "Site A",
+      "connected": true,
+      "rtt": 145338339,
+      "lastSeen": "2025-08-13T14:39:17.208334428-07:00",
+      "endpoint": "p.fosrl.io:21820",
+      "isRelay": true,
+      "peerAddress": "100.89.128.5",
+      "holepunchConnected": false
+    },
+    "8": {
+      "siteId": 8,
+      "name": "Site B",
+      "connected": false,
+      "rtt": 0,
+      "lastSeen": "2025-08-13T14:39:19.663823645-07:00",
+      "endpoint": "p.fosrl.io:21820",
+      "isRelay": true,
+      "peerAddress": "100.89.128.10",
+      "holepunchConnected": false
+    }
+  },
+  "networkSettings": {
+    "tunnelIP": "100.89.128.3/20"
+  }
+}
+```
+
+**Fields:**
+- `connected`: Boolean indicating if connected to Pangolin
+- `registered`: Boolean indicating if registered with the server
+- `terminated`: Boolean indicating if the connection was terminated
+- `version`: Olm version string
+- `agent`: Agent identifier
+- `orgId`: Current organization ID
+- `peers`: Map of peer statuses by site ID
+  - `siteId`: Peer site identifier
+  - `name`: Site name
+  - `connected`: Boolean peer connection state
+  - `rtt`: Peer round-trip time (integer, nanoseconds)
+  - `lastSeen`: Last time peer was seen (RFC3339 timestamp)
+  - `endpoint`: Peer endpoint address
+  - `isRelay`: Whether the peer is relayed (true) or direct (false)
+  - `peerAddress`: Peer's IP address in the tunnel
+  - `holepunchConnected`: Whether holepunch connection is established
+- `networkSettings`: Current network configuration including tunnel IP
+
+**Error Responses:**
+- `405 Method Not Allowed` - Non-GET requests
+
+---
+
+### POST /disconnect
+Disconnects from the current Pangolin server and tears down the WireGuard tunnel.
+
+**Request Body:** None required
+
+**Response:**
+- **Status Code:** `200 OK`
+- **Content-Type:** `application/json`
+
+```json
+{
+  "status": "disconnect initiated"
+}
+```
+
+**Error Responses:**
+- `405 Method Not Allowed` - Non-POST requests
+- `409 Conflict` - Not currently connected to a server
+
+---
+
+### POST /switch-org
+Switches to a different organization while maintaining the connection.
+
+**Request Body:**
+```json
+{
+  "orgId": "string"
+}
+```
+
+**Required Fields:**
+- `orgId`: The organization ID to switch to
+
+**Response:**
+- **Status Code:** `200 OK`
+- **Content-Type:** `application/json`
+
+```json
+{
+  "status": "org switch request accepted"
+}
+```
+
+**Error Responses:**
+- `405 Method Not Allowed` - Non-POST requests
+- `400 Bad Request` - Invalid JSON or missing orgId field
+- `500 Internal Server Error` - Org switch failed
+
+---
+
+### POST /exit
+Initiates a graceful shutdown of the Olm process.
+
+**Request Body:** None required
+
+**Response:**
+- **Status Code:** `200 OK`
+- **Content-Type:** `application/json`
+
+```json
+{
+  "status": "shutdown initiated"
+}
+```
+
+**Note:** The response is sent before shutdown begins. There is a 100ms delay before the actual shutdown to ensure the response is delivered.
+
+**Error Responses:**
+- `405 Method Not Allowed` - Non-POST requests
+
+---
+
+### GET /health
+Simple health check endpoint to verify the API server is running.
+
+**Response:**
+- **Status Code:** `200 OK`
+- **Content-Type:** `application/json`
+
+```json
+{
+  "status": "ok"
+}
+```
+
+**Error Responses:**
+- `405 Method Not Allowed` - Non-GET requests
+
+---
+
+## Usage Examples
+
+### Connect to a peer
+```bash
+curl -X POST http://localhost:9452/connect \
+  -H "Content-Type: application/json" \
+  -d '{
+    "id": "31frd0uzbjvp721",
+    "secret": "h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6",
+    "endpoint": "https://example.com"
+  }'
+```
+
+### Connect with additional options
+```bash
+curl -X POST http://localhost:9452/connect \
+  -H "Content-Type: application/json" \
+  -d '{
+    "id": "31frd0uzbjvp721",
+    "secret": "h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6",
+    "endpoint": "https://example.com",
+    "mtu": 1400,
+    "holepunch": true,
+    "pingInterval": "5s"
+  }'
+```
+
+### Check connection status
+```bash
+curl http://localhost:9452/status
+```
+
+### Switch organization
+```bash
+curl -X POST http://localhost:9452/switch-org \
+  -H "Content-Type: application/json" \
+  -d '{"orgId": "org_456"}'
+```
+
+### Disconnect from server
+```bash
+curl -X POST http://localhost:9452/disconnect
+```
+
+### Health check
+```bash
+curl http://localhost:9452/health
+```
+
+### Shutdown Olm
+```bash
+curl -X POST http://localhost:9452/exit
+```
+
+### Using Unix socket (Linux/macOS)
+```bash
+curl --unix-socket /var/run/olm/olm.sock http://localhost/status
+curl --unix-socket /var/run/olm/olm.sock -X POST http://localhost/disconnect
+```

CONTRIBUTING.md

@@ -4,11 +4,7 @@ Contributions are welcome!
 
 Please see the contribution and local development guide on the docs page before getting started:
 
-https://docs.fossorial.io/development
-
-For ideas about what features to work on and our future plans, please see the roadmap:
-
-https://docs.fossorial.io/roadmap
+https://docs.pangolin.net/development/contributing
 
 ### Licensing Considerations
 

Dockerfile

@@ -1,4 +1,4 @@
-FROM golang:1.23.1-alpine AS builder
+FROM golang:1.25-alpine AS builder
 
 # Set the working directory inside the container
 WORKDIR /app
@@ -13,15 +13,15 @@ RUN go mod download
 COPY . .
 
 # Build the application
-RUN CGO_ENABLED=0 GOOS=linux go build -o /newt
+RUN CGO_ENABLED=0 GOOS=linux go build -o /olm
 
 # Start a new stage from scratch
-FROM ubuntu:22.04 AS runner
+FROM alpine:3.23 AS runner
 
-RUN apt-get update && apt-get install ca-certificates -y  && rm -rf /var/lib/apt/lists/*
+RUN apk --no-cache add ca-certificates
 
 # Copy the pre-built binary file from the previous stage and the entrypoint script
-COPY --from=builder /newt /usr/local/bin/
+COPY --from=builder /olm /usr/local/bin/
 COPY entrypoint.sh /
 
 RUN chmod +x /entrypoint.sh
@@ -30,4 +30,4 @@ RUN chmod +x /entrypoint.sh
 ENTRYPOINT ["/entrypoint.sh"]
 
 # Command to run the executable
-CMD ["newt"]
+CMD ["olm"]

Makefile

@@ -1,17 +1,64 @@
+.PHONY: all local docker-build-release
 
-all: build push
+all: local
 
-build:
-	docker build -t fosrl/newt:latest .
+local:
+	CGO_ENABLED=0 go build -o ./bin/olm
 
-push:
-	docker push fosrl/newt:latest
+docker-build-release:
+	@if [ -z "$(tag)" ]; then \
+		echo "Error: tag is required. Usage: make docker-build-release tag=<tag>"; \
+		exit 1; \
+	fi
+	docker buildx build . \
+		--platform linux/arm/v7,linux/arm64,linux/amd64 \
+		-t fosrl/olm:latest \
+		-t fosrl/olm:$(tag) \
+		-f Dockerfile \
+		--push
 
-test:
-	docker run fosrl/newt:latest
+docker-build-dev:
+	docker buildx build . \
+		--platform linux/arm/v7,linux/arm64,linux/amd64 \
+		-t fosrl/olm:latest \
+		-f Dockerfile
 
-local: 
-	 CGO_ENABLED=0 go build -o newt
+.PHONY: go-build-release \
+        go-build-release-linux-arm64 go-build-release-linux-arm32-v7 \
+        go-build-release-linux-arm32-v6 go-build-release-linux-amd64 \
+        go-build-release-linux-riscv64 go-build-release-darwin-arm64 \
+        go-build-release-darwin-amd64 go-build-release-windows-amd64
 
-clean:
-	rm newt
+go-build-release: \
+    go-build-release-linux-arm64 \
+    go-build-release-linux-arm32-v7 \
+    go-build-release-linux-arm32-v6 \
+    go-build-release-linux-amd64 \
+    go-build-release-linux-riscv64 \
+    go-build-release-darwin-arm64 \
+    go-build-release-darwin-amd64 \
+    go-build-release-windows-amd64 \
+
+go-build-release-linux-arm64:
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -o bin/olm_linux_arm64
+
+go-build-release-linux-arm32-v7:
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm GOARM=7 go build -o bin/olm_linux_arm32
+
+go-build-release-linux-arm32-v6:
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm GOARM=6 go build -o bin/olm_linux_arm32v6
+
+go-build-release-linux-amd64:
+	CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o bin/olm_linux_amd64
+
+go-build-release-linux-riscv64:
+	CGO_ENABLED=0 GOOS=linux GOARCH=riscv64 go build -o bin/olm_linux_riscv64
+
+go-build-release-darwin-arm64:
+	CGO_ENABLED=0 GOOS=darwin GOARCH=arm64 go build -o bin/olm_darwin_arm64
+
+go-build-release-darwin-amd64:
+	CGO_ENABLED=0 GOOS=darwin GOARCH=amd64 go build -o bin/olm_darwin_amd64
+
+go-build-release-windows-amd64:
+	CGO_ENABLED=0 GOOS=windows GOARCH=amd64 go build -o bin/olm_windows_amd64.exe

README.md

@@ -1,101 +1,41 @@
-# Newt
+# Olm
 
-Newt is a fully user space [WireGuard](https://www.wireguard.com/) tunnel client and TCP/UDP proxy, designed to securely expose private resources controlled by Pangolin. By using Newt, you don't need to manage complex WireGuard tunnels and NATing.
+Olm is a [WireGuard](https://www.wireguard.com/) tunnel client designed to securely connect your computer to Newt sites running on remote networks.
 
 ### Installation and Documentation
 
-Newt is used with Pangolin and Gerbil as part of the larger system. See documentation below:
+Olm is used with Pangolin and Newt as part of the larger system. See documentation below:
 
--   [Installation Instructions](https://docs.fossorial.io)
--   [Full Documentation](https://docs.fossorial.io)
-
-## Preview
-
-<img src="public/screenshots/preview.png" alt="Preview"/>
-
-_Sample output of a Newt container connected to Pangolin and hosting various resource target proxies._
+-   [Full Documentation](https://docs.pangolin.net/manage/clients/understanding-clients)
 
 ## Key Functions
 
 ### Registers with Pangolin
 
-Using the Newt ID and a secret the client will make HTTP requests to Pangolin to receive a session token. Using that token it will connect to a websocket and maintain that connection. Control messages will be sent over the websocket.
+Using the Olm ID and a secret, the olm will make HTTP requests to Pangolin to receive a session token. Using that token, it will connect to a websocket and maintain that connection. Control messages will be sent over the websocket.
 
 ### Receives WireGuard Control Messages
 
-When Newt receives WireGuard control messages, it will use the information encoded (endpoint, public key) to bring up a WireGuard tunnel using [netstack](https://github.com/WireGuard/wireguard-go/blob/master/tun/netstack/examples/http_server.go) fully in user space. It will ping over the tunnel to ensure the peer on the Gerbil side is brought up. 
+When Olm receives WireGuard control messages, it will use the information encoded (endpoint, public key) to bring up a WireGuard tunnel on your computer to a remote Newt. It will ping over the tunnel to ensure the peer is brought up.
 
-### Receives Proxy Control Messages
+## Hole Punching
 
-When Newt receives WireGuard control messages, it will use the information encoded to crate local low level TCP and UDP proxies attached to the virtual tunnel in order to relay traffic to programmed targets.
-
-## CLI Args
-
-- `endpoint`: The endpoint where both Gerbil and Pangolin reside in order to connect to the websocket.
-- `id`: Newt ID generated by Pangolin to identify the client.
-- `secret`: A unique secret (not shared and kept private) used to authenticate the client ID with the websocket in order to receive commands. 
-- `dns`: DNS server to use to resolve the endpoint
-- `log-level` (optional): The log level to use. Default: INFO
-
-Example:
-
-```bash
-./newt \
---id 31frd0uzbjvp721 \
---secret h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6 \
---endpoint https://example.com
-```
-
-You can also run it with Docker compose. For example, a service in your `docker-compose.yml` might look like this using environment vars (recommended):
-
-```yaml
-services:
-  newt:
-    image: fosrl/newt
-    container_name: newt
-    restart: unless-stopped
-    environment:
-      - PANGOLIN_ENDPOINT=https://example.com
-      - NEWT_ID=2ix2t8xk22ubpfy 
-      - NEWT_SECRET=nnisrfsdfc7prqsp9ewo1dvtvci50j5uiqotez00dgap0ii2 
-```
-
-You can also pass the CLI args to the container:
-
-```yaml
-services:
-  newt:
-    image: fosrl/newt
-    container_name: newt
-    restart: unless-stopped
-    command:
-        - --id 31frd0uzbjvp721
-        - --secret h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6
-        - --endpoint https://example.com
-```
+In the default mode, olm uses both relaying through Gerbil and NAT hole punching to connect to Newt. Hole punching attempts to orchestrate a NAT traversal between the two sites so that traffic flows directly, which can save data costs and improve speed. If hole punching fails, traffic will fall back to relaying through Gerbil.
 
 ## Build
 
-### Container 
+### Binary
 
-Ensure Docker is installed.
+Make sure to have Go 1.25 installed.
 
 ```bash
 make
 ```
 
-### Binary
-
-Make sure to have Go 1.23.1 installed.
-
-```bash
-make local
-```
-
 ## Licensing
 
-Newt is dual licensed under the AGPLv3 and the Fossorial Commercial license. For inquiries about commercial licensing, please contact us.
+Olm is dual licensed under the AGPLv3 and the Fossorial Commercial license. For inquiries about commercial licensing, please contact us.
 
 ## Contributions
 
-Please see [CONTRIBUTIONS](./CONTRIBUTING.md) in the repository for guidelines and best practices.
+Please see [CONTRIBUTIONS](./CONTRIBUTING.md) in the repository for guidelines and best practices.

SECURITY.md

@@ -0,0 +1,14 @@
+# Security Policy
+
+If you discover a security vulnerability, please follow the steps below to responsibly disclose it to us:
+
+1. **Do not create a public GitHub issue or discussion post.** This could put the security of other users at risk.
+2. Send a detailed report to [security@pangolin.net](mailto:security@pangolin.net) or send a **private** message to a maintainer on [Discord](https://discord.gg/HCJR8Xhme4). Include:
+
+- Description and location of the vulnerability.
+- Potential impact of the vulnerability.
+- Steps to reproduce the vulnerability.
+- Potential solutions to fix the vulnerability.
+- Your name/handle and a link for recognition (optional).
+
+We aim to address the issue as soon as possible.

api/api.go

@@ -0,0 +1,503 @@
+package api
+
+import (
+	"encoding/json"
+	"fmt"
+	"net"
+	"net/http"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/network"
+)
+
+// ConnectionRequest defines the structure for an incoming connection request
+type ConnectionRequest struct {
+	ID            string   `json:"id"`
+	Secret        string   `json:"secret"`
+	Endpoint      string   `json:"endpoint"`
+	UserToken     string   `json:"userToken,omitempty"`
+	MTU           int      `json:"mtu,omitempty"`
+	DNS           string   `json:"dns,omitempty"`
+	DNSProxyIP    string   `json:"dnsProxyIP,omitempty"`
+	UpstreamDNS   []string `json:"upstreamDNS,omitempty"`
+	InterfaceName string   `json:"interfaceName,omitempty"`
+	Holepunch     bool     `json:"holepunch,omitempty"`
+	TlsClientCert string   `json:"tlsClientCert,omitempty"`
+	PingInterval  string   `json:"pingInterval,omitempty"`
+	PingTimeout   string   `json:"pingTimeout,omitempty"`
+	OrgID         string   `json:"orgId,omitempty"`
+}
+
+// SwitchOrgRequest defines the structure for switching organizations
+type SwitchOrgRequest struct {
+	OrgID string `json:"orgId"`
+}
+
+// PeerStatus represents the status of a peer connection
+type PeerStatus struct {
+	SiteID             int           `json:"siteId"`
+	Name               string        `json:"name"`
+	Connected          bool          `json:"connected"`
+	RTT                time.Duration `json:"rtt"`
+	LastSeen           time.Time     `json:"lastSeen"`
+	Endpoint           string        `json:"endpoint,omitempty"`
+	IsRelay            bool          `json:"isRelay"`
+	PeerIP             string        `json:"peerAddress,omitempty"`
+	HolepunchConnected bool          `json:"holepunchConnected"`
+}
+
+// StatusResponse is returned by the status endpoint
+type StatusResponse struct {
+	Connected       bool                    `json:"connected"`
+	Registered      bool                    `json:"registered"`
+	Terminated      bool                    `json:"terminated"`
+	Version         string                  `json:"version,omitempty"`
+	Agent           string                  `json:"agent,omitempty"`
+	OrgID           string                  `json:"orgId,omitempty"`
+	PeerStatuses    map[int]*PeerStatus     `json:"peers,omitempty"`
+	NetworkSettings network.NetworkSettings `json:"networkSettings,omitempty"`
+}
+
+// API represents the HTTP server and its state
+type API struct {
+	addr         string
+	socketPath   string
+	listener     net.Listener
+	server       *http.Server
+	onConnect    func(ConnectionRequest) error
+	onSwitchOrg  func(SwitchOrgRequest) error
+	onDisconnect func() error
+	onExit       func() error
+	statusMu     sync.RWMutex
+	peerStatuses map[int]*PeerStatus
+	connectedAt  time.Time
+	isConnected  bool
+	isRegistered bool
+	isTerminated bool
+	version      string
+	agent        string
+	orgID        string
+}
+
+// NewAPI creates a new HTTP server that listens on a TCP address
+func NewAPI(addr string) *API {
+	s := &API{
+		addr:         addr,
+		peerStatuses: make(map[int]*PeerStatus),
+	}
+
+	return s
+}
+
+// NewAPISocket creates a new HTTP server that listens on a Unix socket or Windows named pipe
+func NewAPISocket(socketPath string) *API {
+	s := &API{
+		socketPath:   socketPath,
+		peerStatuses: make(map[int]*PeerStatus),
+	}
+
+	return s
+}
+
+// SetHandlers sets the callback functions for handling API requests
+func (s *API) SetHandlers(
+	onConnect func(ConnectionRequest) error,
+	onSwitchOrg func(SwitchOrgRequest) error,
+	onDisconnect func() error,
+	onExit func() error,
+) {
+	s.onConnect = onConnect
+	s.onSwitchOrg = onSwitchOrg
+	s.onDisconnect = onDisconnect
+	s.onExit = onExit
+}
+
+// Start starts the HTTP server
+func (s *API) Start() error {
+	mux := http.NewServeMux()
+	mux.HandleFunc("/connect", s.handleConnect)
+	mux.HandleFunc("/status", s.handleStatus)
+	mux.HandleFunc("/switch-org", s.handleSwitchOrg)
+	mux.HandleFunc("/disconnect", s.handleDisconnect)
+	mux.HandleFunc("/exit", s.handleExit)
+	mux.HandleFunc("/health", s.handleHealth)
+
+	s.server = &http.Server{
+		Handler: mux,
+	}
+
+	var err error
+	if s.socketPath != "" {
+		// Use platform-specific socket listener
+		s.listener, err = createSocketListener(s.socketPath)
+		if err != nil {
+			return fmt.Errorf("failed to create socket listener: %w", err)
+		}
+		logger.Info("Starting HTTP server on socket %s", s.socketPath)
+	} else {
+		// Use TCP listener
+		s.listener, err = net.Listen("tcp", s.addr)
+		if err != nil {
+			return fmt.Errorf("failed to create TCP listener: %w", err)
+		}
+		logger.Info("Starting HTTP server on %s", s.addr)
+	}
+
+	go func() {
+		if err := s.server.Serve(s.listener); err != nil && err != http.ErrServerClosed {
+			logger.Error("HTTP server error: %v", err)
+		}
+	}()
+
+	return nil
+}
+
+// Stop stops the HTTP server
+func (s *API) Stop() error {
+	logger.Info("Stopping api server")
+
+	// Close the server first, which will also close the listener gracefully
+	if s.server != nil {
+		s.server.Close()
+	}
+
+	// Clean up socket file if using Unix socket
+	if s.socketPath != "" {
+		cleanupSocket(s.socketPath)
+	}
+
+	return nil
+}
+
+func (s *API) AddPeerStatus(siteID int, siteName string, connected bool, rtt time.Duration, endpoint string, isRelay bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+
+	status, exists := s.peerStatuses[siteID]
+	if !exists {
+		status = &PeerStatus{
+			SiteID: siteID,
+		}
+		s.peerStatuses[siteID] = status
+	}
+
+	status.Name = siteName
+	status.Connected = connected
+	status.RTT = rtt
+	status.LastSeen = time.Now()
+	status.Endpoint = endpoint
+	status.IsRelay = isRelay
+}
+
+// UpdatePeerStatus updates the status of a peer including endpoint and relay info
+func (s *API) UpdatePeerStatus(siteID int, connected bool, rtt time.Duration, endpoint string, isRelay bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+
+	status, exists := s.peerStatuses[siteID]
+	if !exists {
+		status = &PeerStatus{
+			SiteID: siteID,
+		}
+		s.peerStatuses[siteID] = status
+	}
+
+	status.Connected = connected
+	status.RTT = rtt
+	status.LastSeen = time.Now()
+	status.Endpoint = endpoint
+	status.IsRelay = isRelay
+}
+
+func (s *API) RemovePeerStatus(siteID int) { // remove the peer from the status map
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	delete(s.peerStatuses, siteID)
+}
+
+// SetConnectionStatus sets the overall connection status
+func (s *API) SetConnectionStatus(isConnected bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+
+	s.isConnected = isConnected
+
+	if isConnected {
+		s.connectedAt = time.Now()
+	} else {
+		// Clear peer statuses when disconnected
+		s.peerStatuses = make(map[int]*PeerStatus)
+	}
+}
+
+func (s *API) SetRegistered(registered bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	s.isRegistered = registered
+}
+
+func (s *API) SetTerminated(terminated bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	s.isTerminated = terminated
+}
+
+// ClearPeerStatuses clears all peer statuses
+func (s *API) ClearPeerStatuses() {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	s.peerStatuses = make(map[int]*PeerStatus)
+}
+
+// SetVersion sets the olm version
+func (s *API) SetVersion(version string) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	s.version = version
+}
+
+// SetAgent sets the olm agent
+func (s *API) SetAgent(agent string) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	s.agent = agent
+}
+
+// SetOrgID sets the organization ID
+func (s *API) SetOrgID(orgID string) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+	s.orgID = orgID
+}
+
+// UpdatePeerRelayStatus updates only the relay status of a peer
+func (s *API) UpdatePeerRelayStatus(siteID int, endpoint string, isRelay bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+
+	status, exists := s.peerStatuses[siteID]
+	if !exists {
+		status = &PeerStatus{
+			SiteID: siteID,
+		}
+		s.peerStatuses[siteID] = status
+	}
+
+	status.Endpoint = endpoint
+	status.IsRelay = isRelay
+}
+
+// UpdatePeerHolepunchStatus updates the holepunch connection status of a peer
+func (s *API) UpdatePeerHolepunchStatus(siteID int, holepunchConnected bool) {
+	s.statusMu.Lock()
+	defer s.statusMu.Unlock()
+
+	status, exists := s.peerStatuses[siteID]
+	if !exists {
+		status = &PeerStatus{
+			SiteID: siteID,
+		}
+		s.peerStatuses[siteID] = status
+	}
+
+	status.HolepunchConnected = holepunchConnected
+}
+
+// handleConnect handles the /connect endpoint
+func (s *API) handleConnect(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+
+	// if we are already connected, reject new connection requests
+	s.statusMu.RLock()
+	alreadyConnected := s.isConnected
+	s.statusMu.RUnlock()
+	if alreadyConnected {
+		http.Error(w, "Already connected to a server. Disconnect first before connecting again.", http.StatusConflict)
+		return
+	}
+
+	var req ConnectionRequest
+	decoder := json.NewDecoder(r.Body)
+	if err := decoder.Decode(&req); err != nil {
+		http.Error(w, fmt.Sprintf("Invalid request: %v", err), http.StatusBadRequest)
+		return
+	}
+
+	// Validate required fields
+	if req.ID == "" || req.Secret == "" || req.Endpoint == "" {
+		http.Error(w, "Missing required fields: id, secret, and endpoint must be provided", http.StatusBadRequest)
+		return
+	}
+
+	// Call the connect handler if set
+	if s.onConnect != nil {
+		if err := s.onConnect(req); err != nil {
+			http.Error(w, fmt.Sprintf("Connection failed: %v", err), http.StatusInternalServerError)
+			return
+		}
+	}
+
+	// Return a success response
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(http.StatusAccepted)
+	json.NewEncoder(w).Encode(map[string]string{
+		"status": "connection request accepted",
+	})
+}
+
+// handleStatus handles the /status endpoint
+func (s *API) handleStatus(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodGet {
+		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+
+	s.statusMu.RLock()
+	defer s.statusMu.RUnlock()
+
+	resp := StatusResponse{
+		Connected:       s.isConnected,
+		Registered:      s.isRegistered,
+		Terminated:      s.isTerminated,
+		Version:         s.version,
+		Agent:           s.agent,
+		OrgID:           s.orgID,
+		PeerStatuses:    s.peerStatuses,
+		NetworkSettings: network.GetSettings(),
+	}
+
+	w.Header().Set("Content-Type", "application/json")
+	json.NewEncoder(w).Encode(resp)
+}
+
+// handleHealth handles the /health endpoint
+func (s *API) handleHealth(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodGet {
+		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(http.StatusOK)
+	json.NewEncoder(w).Encode(map[string]string{
+		"status": "ok",
+	})
+}
+
+// handleExit handles the /exit endpoint
+func (s *API) handleExit(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+
+	logger.Info("Received exit request via API")
+
+	// Return a success response first
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(http.StatusOK)
+	json.NewEncoder(w).Encode(map[string]string{
+		"status": "shutdown initiated",
+	})
+
+	// Call the exit handler after responding, in a goroutine with a small delay
+	// to ensure the response is fully sent before shutdown begins
+	if s.onExit != nil {
+		go func() {
+			time.Sleep(100 * time.Millisecond)
+			if err := s.onExit(); err != nil {
+				logger.Error("Exit handler failed: %v", err)
+			}
+		}()
+	}
+}
+
+// handleSwitchOrg handles the /switch-org endpoint
+func (s *API) handleSwitchOrg(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+
+	var req SwitchOrgRequest
+	decoder := json.NewDecoder(r.Body)
+	if err := decoder.Decode(&req); err != nil {
+		http.Error(w, fmt.Sprintf("Invalid request: %v", err), http.StatusBadRequest)
+		return
+	}
+
+	// Validate required fields
+	if req.OrgID == "" {
+		http.Error(w, "Missing required field: orgId must be provided", http.StatusBadRequest)
+		return
+	}
+
+	logger.Info("Received org switch request to orgId: %s", req.OrgID)
+
+	// Call the switch org handler if set
+	if s.onSwitchOrg != nil {
+		if err := s.onSwitchOrg(req); err != nil {
+			http.Error(w, fmt.Sprintf("Org switch failed: %v", err), http.StatusInternalServerError)
+			return
+		}
+	}
+
+	// Return a success response
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(http.StatusOK)
+	json.NewEncoder(w).Encode(map[string]string{
+		"status": "org switch request accepted",
+	})
+}
+
+// handleDisconnect handles the /disconnect endpoint
+func (s *API) handleDisconnect(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
+		return
+	}
+
+	// if we are already disconnected, reject new disconnect requests
+	s.statusMu.RLock()
+	alreadyDisconnected := !s.isConnected
+	s.statusMu.RUnlock()
+	if alreadyDisconnected {
+		http.Error(w, "Not currently connected to a server.", http.StatusConflict)
+		return
+	}
+
+	logger.Info("Received disconnect request via API")
+
+	// Call the disconnect handler if set
+	if s.onDisconnect != nil {
+		if err := s.onDisconnect(); err != nil {
+			http.Error(w, fmt.Sprintf("Disconnect failed: %v", err), http.StatusInternalServerError)
+			return
+		}
+	}
+
+	// Return a success response
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(http.StatusOK)
+	json.NewEncoder(w).Encode(map[string]string{
+		"status": "disconnect initiated",
+	})
+}
+
+func (s *API) GetStatus() StatusResponse {
+	return StatusResponse{
+		Connected:       s.isConnected,
+		Registered:      s.isRegistered,
+		Terminated:      s.isTerminated,
+		Version:         s.version,
+		Agent:           s.agent,
+		OrgID:           s.orgID,
+		PeerStatuses:    s.peerStatuses,
+		NetworkSettings: network.GetSettings(),
+	}
+}

api/api_unix.go

@@ -0,0 +1,50 @@
+//go:build !windows
+// +build !windows
+
+package api
+
+import (
+	"fmt"
+	"net"
+	"os"
+	"path/filepath"
+
+	"github.com/fosrl/newt/logger"
+)
+
+// createSocketListener creates a Unix domain socket listener
+func createSocketListener(socketPath string) (net.Listener, error) {
+	// Ensure the directory exists
+	dir := filepath.Dir(socketPath)
+	if err := os.MkdirAll(dir, 0755); err != nil {
+		return nil, fmt.Errorf("failed to create socket directory: %w", err)
+	}
+
+	// Remove existing socket file if it exists
+	if err := os.RemoveAll(socketPath); err != nil {
+		return nil, fmt.Errorf("failed to remove existing socket: %w", err)
+	}
+
+	listener, err := net.Listen("unix", socketPath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to listen on Unix socket: %w", err)
+	}
+
+	// Set socket permissions to allow access
+	if err := os.Chmod(socketPath, 0666); err != nil {
+		listener.Close()
+		return nil, fmt.Errorf("failed to set socket permissions: %w", err)
+	}
+
+	logger.Debug("Created Unix socket at %s", socketPath)
+	return listener, nil
+}
+
+// cleanupSocket removes the Unix socket file
+func cleanupSocket(socketPath string) {
+	if err := os.Remove(socketPath); err != nil && !os.IsNotExist(err) {
+		logger.Error("Failed to remove socket file %s: %v", socketPath, err)
+	} else {
+		logger.Debug("Removed Unix socket at %s", socketPath)
+	}
+}

api/api_windows.go

@@ -0,0 +1,41 @@
+//go:build windows
+// +build windows
+
+package api
+
+import (
+	"fmt"
+	"net"
+
+	"github.com/Microsoft/go-winio"
+	"github.com/fosrl/newt/logger"
+)
+
+// createSocketListener creates a Windows named pipe listener
+func createSocketListener(pipePath string) (net.Listener, error) {
+	// Ensure the pipe path has the correct format
+	if pipePath[0] != '\\' {
+		pipePath = `\\.\pipe\` + pipePath
+	}
+
+	// Create a pipe configuration that allows everyone to write
+	config := &winio.PipeConfig{
+		// Set security descriptor to allow everyone full access
+		// This SDDL string grants full access to Everyone (WD) and to the current owner (OW)
+		SecurityDescriptor: "D:(A;;GA;;;WD)(A;;GA;;;OW)",
+	}
+
+	// Create a named pipe listener using go-winio with the configuration
+	listener, err := winio.ListenPipe(pipePath, config)
+	if err != nil {
+		return nil, fmt.Errorf("failed to listen on named pipe: %w", err)
+	}
+
+	logger.Debug("Created named pipe at %s with write access for everyone", pipePath)
+	return listener, nil
+}
+
+// cleanupSocket is a no-op on Windows as named pipes are automatically cleaned up
+func cleanupSocket(pipePath string) {
+	logger.Debug("Named pipe %s will be automatically cleaned up", pipePath)
+}

config.go

@@ -0,0 +1,649 @@
+package main
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"os"
+	"path/filepath"
+	"runtime"
+	"strconv"
+	"strings"
+	"time"
+)
+
+// OlmConfig holds all configuration options for the Olm client
+type OlmConfig struct {
+	// Connection settings
+	Endpoint  string `json:"endpoint"`
+	ID        string `json:"id"`
+	Secret    string `json:"secret"`
+	OrgID     string `json:"org"`
+	UserToken string `json:"userToken"`
+
+	// Network settings
+	MTU           int      `json:"mtu"`
+	DNS           string   `json:"dns"`
+	UpstreamDNS   []string `json:"upstreamDNS"`
+	InterfaceName string   `json:"interface"`
+
+	// Logging
+	LogLevel string `json:"logLevel"`
+
+	// HTTP server
+	EnableAPI  bool   `json:"enableApi"`
+	HTTPAddr   string `json:"httpAddr"`
+	SocketPath string `json:"socketPath"`
+
+	// Ping settings
+	PingInterval string `json:"pingInterval"`
+	PingTimeout  string `json:"pingTimeout"`
+
+	// Advanced
+	DisableHolepunch bool   `json:"disableHolepunch"`
+	TlsClientCert    string `json:"tlsClientCert"`
+	OverrideDNS      bool   `json:"overrideDNS"`
+	TunnelDNS        bool   `json:"tunnelDNS"`
+	DisableRelay     bool   `json:"disableRelay"`
+	// DoNotCreateNewClient bool   `json:"doNotCreateNewClient"`
+
+	// Parsed values (not in JSON)
+	PingIntervalDuration time.Duration `json:"-"`
+	PingTimeoutDuration  time.Duration `json:"-"`
+
+	// Source tracking (not in JSON)
+	sources map[string]string `json:"-"`
+
+	Version string
+}
+
+// ConfigSource tracks where each config value came from
+type ConfigSource string
+
+const (
+	SourceDefault ConfigSource = "default"
+	SourceFile    ConfigSource = "file"
+	SourceEnv     ConfigSource = "environment"
+	SourceCLI     ConfigSource = "cli"
+)
+
+// DefaultConfig returns a config with default values
+func DefaultConfig() *OlmConfig {
+	// Set OS-specific socket path
+	var socketPath string
+	switch runtime.GOOS {
+	case "windows":
+		socketPath = "olm"
+	default: // darwin, linux, and others
+		socketPath = "/var/run/olm.sock"
+	}
+
+	config := &OlmConfig{
+		MTU:              1280,
+		DNS:              "8.8.8.8",
+		UpstreamDNS:      []string{"8.8.8.8:53"},
+		LogLevel:         "INFO",
+		InterfaceName:    "olm",
+		EnableAPI:        false,
+		SocketPath:       socketPath,
+		PingInterval:     "3s",
+		PingTimeout:      "5s",
+		DisableHolepunch: false,
+		TunnelDNS:        false,
+		// DoNotCreateNewClient: false,
+		sources: make(map[string]string),
+	}
+
+	// Track default sources
+	config.sources["mtu"] = string(SourceDefault)
+	config.sources["dns"] = string(SourceDefault)
+	config.sources["upstreamDNS"] = string(SourceDefault)
+	config.sources["logLevel"] = string(SourceDefault)
+	config.sources["interface"] = string(SourceDefault)
+	config.sources["enableApi"] = string(SourceDefault)
+	config.sources["httpAddr"] = string(SourceDefault)
+	config.sources["socketPath"] = string(SourceDefault)
+	config.sources["pingInterval"] = string(SourceDefault)
+	config.sources["pingTimeout"] = string(SourceDefault)
+	config.sources["disableHolepunch"] = string(SourceDefault)
+	config.sources["overrideDNS"] = string(SourceDefault)
+	config.sources["tunnelDNS"] = string(SourceDefault)
+	config.sources["disableRelay"] = string(SourceDefault)
+	// config.sources["doNotCreateNewClient"] = string(SourceDefault)
+
+	return config
+}
+
+// getOlmConfigPath returns the path to the olm config file
+func getOlmConfigPath() string {
+	configFile := os.Getenv("CONFIG_FILE")
+	if configFile != "" {
+		return configFile
+	}
+
+	var configDir string
+	switch runtime.GOOS {
+	case "darwin":
+		configDir = filepath.Join(os.Getenv("HOME"), "Library", "Application Support", "olm-client")
+	case "windows":
+		configDir = filepath.Join(os.Getenv("PROGRAMDATA"), "olm", "olm-client")
+	default: // linux and others
+		configDir = filepath.Join(os.Getenv("HOME"), ".config", "olm-client")
+	}
+
+	if err := os.MkdirAll(configDir, 0755); err != nil {
+		fmt.Printf("Warning: Failed to create config directory: %v\n", err)
+	}
+
+	return filepath.Join(configDir, "config.json")
+}
+
+// LoadConfig loads configuration from file, env vars, and CLI args
+// Priority: CLI args > Env vars > Config file > Defaults
+// Returns: (config, showVersion, showConfig, error)
+func LoadConfig(args []string) (*OlmConfig, bool, bool, error) {
+	// Start with defaults
+	config := DefaultConfig()
+
+	// Load from config file (if exists)
+	fileConfig, err := loadConfigFromFile()
+	if err != nil {
+		return nil, false, false, fmt.Errorf("failed to load config file: %w", err)
+	}
+	if fileConfig != nil {
+		mergeConfigs(config, fileConfig)
+	}
+
+	// Override with environment variables
+	loadConfigFromEnv(config)
+
+	// Override with CLI arguments
+	showVersion, showConfig, err := loadConfigFromCLI(config, args)
+	if err != nil {
+		return nil, false, false, err
+	}
+
+	// Parse duration strings
+	if err := config.parseDurations(); err != nil {
+		return nil, false, false, err
+	}
+
+	return config, showVersion, showConfig, nil
+}
+
+// loadConfigFromFile loads configuration from the JSON config file
+func loadConfigFromFile() (*OlmConfig, error) {
+	configPath := getOlmConfigPath()
+	data, err := os.ReadFile(configPath)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil, nil // File doesn't exist, not an error
+		}
+		return nil, err
+	}
+
+	var config OlmConfig
+	if err := json.Unmarshal(data, &config); err != nil {
+		return nil, fmt.Errorf("failed to parse config file: %w", err)
+	}
+
+	return &config, nil
+}
+
+// loadConfigFromEnv loads configuration from environment variables
+func loadConfigFromEnv(config *OlmConfig) {
+	if val := os.Getenv("PANGOLIN_ENDPOINT"); val != "" {
+		config.Endpoint = val
+		config.sources["endpoint"] = string(SourceEnv)
+	}
+	if val := os.Getenv("OLM_ID"); val != "" {
+		config.ID = val
+		config.sources["id"] = string(SourceEnv)
+	}
+	if val := os.Getenv("OLM_SECRET"); val != "" {
+		config.Secret = val
+		config.sources["secret"] = string(SourceEnv)
+	}
+	if val := os.Getenv("ORG"); val != "" {
+		config.OrgID = val
+		config.sources["org"] = string(SourceEnv)
+	}
+	if val := os.Getenv("USER_TOKEN"); val != "" {
+		config.UserToken = val
+		config.sources["userToken"] = string(SourceEnv)
+	}
+	if val := os.Getenv("MTU"); val != "" {
+		if mtu, err := strconv.Atoi(val); err == nil {
+			config.MTU = mtu
+			config.sources["mtu"] = string(SourceEnv)
+		} else {
+			fmt.Printf("Invalid MTU value: %s, keeping current value\n", val)
+		}
+	}
+	if val := os.Getenv("DNS"); val != "" {
+		config.DNS = val
+		config.sources["dns"] = string(SourceEnv)
+	}
+	if val := os.Getenv("UPSTREAM_DNS"); val != "" {
+		config.UpstreamDNS = []string{val}
+		config.sources["upstreamDNS"] = string(SourceEnv)
+	}
+	if val := os.Getenv("LOG_LEVEL"); val != "" {
+		config.LogLevel = val
+		config.sources["logLevel"] = string(SourceEnv)
+	}
+	if val := os.Getenv("INTERFACE"); val != "" {
+		config.InterfaceName = val
+		config.sources["interface"] = string(SourceEnv)
+	}
+	if val := os.Getenv("HTTP_ADDR"); val != "" {
+		config.HTTPAddr = val
+		config.sources["httpAddr"] = string(SourceEnv)
+	}
+	if val := os.Getenv("PING_INTERVAL"); val != "" {
+		config.PingInterval = val
+		config.sources["pingInterval"] = string(SourceEnv)
+	}
+	if val := os.Getenv("PING_TIMEOUT"); val != "" {
+		config.PingTimeout = val
+		config.sources["pingTimeout"] = string(SourceEnv)
+	}
+	if val := os.Getenv("ENABLE_API"); val == "true" {
+		config.EnableAPI = true
+		config.sources["enableApi"] = string(SourceEnv)
+	}
+	if val := os.Getenv("SOCKET_PATH"); val != "" {
+		config.SocketPath = val
+		config.sources["socketPath"] = string(SourceEnv)
+	}
+	if val := os.Getenv("DISABLE_HOLEPUNCH"); val == "true" {
+		config.DisableHolepunch = true
+		config.sources["disableHolepunch"] = string(SourceEnv)
+	}
+	if val := os.Getenv("OVERRIDE_DNS"); val == "true" {
+		config.OverrideDNS = true
+		config.sources["overrideDNS"] = string(SourceEnv)
+	}
+	if val := os.Getenv("DISABLE_RELAY"); val == "true" {
+		config.DisableRelay = true
+		config.sources["disableRelay"] = string(SourceEnv)
+	}
+	if val := os.Getenv("TUNNEL_DNS"); val == "true" {
+		config.TunnelDNS = true
+		config.sources["tunnelDNS"] = string(SourceEnv)
+	}
+	// if val := os.Getenv("DO_NOT_CREATE_NEW_CLIENT"); val == "true" {
+	// 	config.DoNotCreateNewClient = true
+	// 	config.sources["doNotCreateNewClient"] = string(SourceEnv)
+	// }
+}
+
+// loadConfigFromCLI loads configuration from command-line arguments
+func loadConfigFromCLI(config *OlmConfig, args []string) (bool, bool, error) {
+	serviceFlags := flag.NewFlagSet("service", flag.ContinueOnError)
+
+	// Store original values to detect changes
+	origValues := map[string]interface{}{
+		"endpoint":         config.Endpoint,
+		"id":               config.ID,
+		"secret":           config.Secret,
+		"org":              config.OrgID,
+		"userToken":        config.UserToken,
+		"mtu":              config.MTU,
+		"dns":              config.DNS,
+		"upstreamDNS":      fmt.Sprintf("%v", config.UpstreamDNS),
+		"logLevel":         config.LogLevel,
+		"interface":        config.InterfaceName,
+		"httpAddr":         config.HTTPAddr,
+		"socketPath":       config.SocketPath,
+		"pingInterval":     config.PingInterval,
+		"pingTimeout":      config.PingTimeout,
+		"enableApi":        config.EnableAPI,
+		"disableHolepunch": config.DisableHolepunch,
+		"overrideDNS":      config.OverrideDNS,
+		"disableRelay":     config.DisableRelay,
+		"tunnelDNS":        config.TunnelDNS,
+		// "doNotCreateNewClient": config.DoNotCreateNewClient,
+	}
+
+	// Define flags
+	serviceFlags.StringVar(&config.Endpoint, "endpoint", config.Endpoint, "Endpoint of your Pangolin server")
+	serviceFlags.StringVar(&config.ID, "id", config.ID, "Olm ID")
+	serviceFlags.StringVar(&config.Secret, "secret", config.Secret, "Olm secret")
+	serviceFlags.StringVar(&config.OrgID, "org", config.OrgID, "Organization ID")
+	serviceFlags.StringVar(&config.UserToken, "user-token", config.UserToken, "User token (optional)")
+	serviceFlags.IntVar(&config.MTU, "mtu", config.MTU, "MTU to use")
+	serviceFlags.StringVar(&config.DNS, "dns", config.DNS, "DNS server to use")
+	var upstreamDNSFlag string
+	serviceFlags.StringVar(&upstreamDNSFlag, "upstream-dns", "", "Upstream DNS server(s) (comma-separated, default: 8.8.8.8:53)")
+	serviceFlags.StringVar(&config.LogLevel, "log-level", config.LogLevel, "Log level (DEBUG, INFO, WARN, ERROR, FATAL)")
+	serviceFlags.StringVar(&config.InterfaceName, "interface", config.InterfaceName, "Name of the WireGuard interface")
+	serviceFlags.StringVar(&config.HTTPAddr, "http-addr", config.HTTPAddr, "HTTP server address (e.g., ':9452')")
+	serviceFlags.StringVar(&config.SocketPath, "socket-path", config.SocketPath, "Unix socket path (or named pipe on Windows)")
+	serviceFlags.StringVar(&config.PingInterval, "ping-interval", config.PingInterval, "Interval for pinging the server")
+	serviceFlags.StringVar(&config.PingTimeout, "ping-timeout", config.PingTimeout, "Timeout for each ping")
+	serviceFlags.BoolVar(&config.EnableAPI, "enable-api", config.EnableAPI, "Enable API server for receiving connection requests")
+	serviceFlags.BoolVar(&config.DisableHolepunch, "disable-holepunch", config.DisableHolepunch, "Disable hole punching")
+	serviceFlags.BoolVar(&config.OverrideDNS, "override-dns", config.OverrideDNS, "Override system DNS settings")
+	serviceFlags.BoolVar(&config.DisableRelay, "disable-relay", config.DisableRelay, "Disable relay connections")
+	serviceFlags.BoolVar(&config.TunnelDNS, "tunnel-dns", config.TunnelDNS, "Use tunnel for DNS traffic")
+	// serviceFlags.BoolVar(&config.DoNotCreateNewClient, "do-not-create-new-client", config.DoNotCreateNewClient, "Do not create new client")
+
+	version := serviceFlags.Bool("version", false, "Print the version")
+	showConfig := serviceFlags.Bool("show-config", false, "Show configuration sources and exit")
+
+	// Parse the arguments
+	if err := serviceFlags.Parse(args); err != nil {
+		return false, false, err
+	}
+
+	// Parse upstream DNS flag if provided
+	if upstreamDNSFlag != "" {
+		config.UpstreamDNS = []string{}
+		for _, dns := range splitComma(upstreamDNSFlag) {
+			if dns != "" {
+				config.UpstreamDNS = append(config.UpstreamDNS, dns)
+			}
+		}
+	}
+
+	// Track which values were changed by CLI args
+	if config.Endpoint != origValues["endpoint"].(string) {
+		config.sources["endpoint"] = string(SourceCLI)
+	}
+	if config.ID != origValues["id"].(string) {
+		config.sources["id"] = string(SourceCLI)
+	}
+	if config.Secret != origValues["secret"].(string) {
+		config.sources["secret"] = string(SourceCLI)
+	}
+	if config.OrgID != origValues["org"].(string) {
+		config.sources["org"] = string(SourceCLI)
+	}
+	if config.UserToken != origValues["userToken"].(string) {
+		config.sources["userToken"] = string(SourceCLI)
+	}
+	if config.MTU != origValues["mtu"].(int) {
+		config.sources["mtu"] = string(SourceCLI)
+	}
+	if config.DNS != origValues["dns"].(string) {
+		config.sources["dns"] = string(SourceCLI)
+	}
+	if fmt.Sprintf("%v", config.UpstreamDNS) != origValues["upstreamDNS"].(string) {
+		config.sources["upstreamDNS"] = string(SourceCLI)
+	}
+	if config.LogLevel != origValues["logLevel"].(string) {
+		config.sources["logLevel"] = string(SourceCLI)
+	}
+	if config.InterfaceName != origValues["interface"].(string) {
+		config.sources["interface"] = string(SourceCLI)
+	}
+	if config.HTTPAddr != origValues["httpAddr"].(string) {
+		config.sources["httpAddr"] = string(SourceCLI)
+	}
+	if config.SocketPath != origValues["socketPath"].(string) {
+		config.sources["socketPath"] = string(SourceCLI)
+	}
+	if config.PingInterval != origValues["pingInterval"].(string) {
+		config.sources["pingInterval"] = string(SourceCLI)
+	}
+	if config.PingTimeout != origValues["pingTimeout"].(string) {
+		config.sources["pingTimeout"] = string(SourceCLI)
+	}
+	if config.EnableAPI != origValues["enableApi"].(bool) {
+		config.sources["enableApi"] = string(SourceCLI)
+	}
+	if config.DisableHolepunch != origValues["disableHolepunch"].(bool) {
+		config.sources["disableHolepunch"] = string(SourceCLI)
+	}
+	if config.OverrideDNS != origValues["overrideDNS"].(bool) {
+		config.sources["overrideDNS"] = string(SourceCLI)
+	}
+	if config.DisableRelay != origValues["disableRelay"].(bool) {
+		config.sources["disableRelay"] = string(SourceCLI)
+	}
+	if config.TunnelDNS != origValues["tunnelDNS"].(bool) {
+		config.sources["tunnelDNS"] = string(SourceCLI)
+	}
+	// if config.DoNotCreateNewClient != origValues["doNotCreateNewClient"].(bool) {
+	// 	config.sources["doNotCreateNewClient"] = string(SourceCLI)
+	// }
+
+	return *version, *showConfig, nil
+}
+
+// parseDurations parses the duration strings into time.Duration
+func (c *OlmConfig) parseDurations() error {
+	var err error
+
+	// Parse ping interval
+	if c.PingInterval != "" {
+		c.PingIntervalDuration, err = time.ParseDuration(c.PingInterval)
+		if err != nil {
+			fmt.Printf("Invalid PING_INTERVAL value: %s, using default 3 seconds\n", c.PingInterval)
+			c.PingIntervalDuration = 3 * time.Second
+			c.PingInterval = "3s"
+		}
+	} else {
+		c.PingIntervalDuration = 3 * time.Second
+		c.PingInterval = "3s"
+	}
+
+	// Parse ping timeout
+	if c.PingTimeout != "" {
+		c.PingTimeoutDuration, err = time.ParseDuration(c.PingTimeout)
+		if err != nil {
+			fmt.Printf("Invalid PING_TIMEOUT value: %s, using default 5 seconds\n", c.PingTimeout)
+			c.PingTimeoutDuration = 5 * time.Second
+			c.PingTimeout = "5s"
+		}
+	} else {
+		c.PingTimeoutDuration = 5 * time.Second
+		c.PingTimeout = "5s"
+	}
+
+	return nil
+}
+
+// mergeConfigs merges source config into destination (only non-empty values)
+// Also tracks that these values came from a file
+func mergeConfigs(dest, src *OlmConfig) {
+	if src.Endpoint != "" {
+		dest.Endpoint = src.Endpoint
+		dest.sources["endpoint"] = string(SourceFile)
+	}
+	if src.ID != "" {
+		dest.ID = src.ID
+		dest.sources["id"] = string(SourceFile)
+	}
+	if src.Secret != "" {
+		dest.Secret = src.Secret
+		dest.sources["secret"] = string(SourceFile)
+	}
+	if src.OrgID != "" {
+		dest.OrgID = src.OrgID
+		dest.sources["org"] = string(SourceFile)
+	}
+	if src.UserToken != "" {
+		dest.UserToken = src.UserToken
+		dest.sources["userToken"] = string(SourceFile)
+	}
+	if src.MTU != 0 && src.MTU != 1280 {
+		dest.MTU = src.MTU
+		dest.sources["mtu"] = string(SourceFile)
+	}
+	if src.DNS != "" && src.DNS != "8.8.8.8" {
+		dest.DNS = src.DNS
+		dest.sources["dns"] = string(SourceFile)
+	}
+	if len(src.UpstreamDNS) > 0 && fmt.Sprintf("%v", src.UpstreamDNS) != "[8.8.8.8:53]" {
+		dest.UpstreamDNS = src.UpstreamDNS
+		dest.sources["upstreamDNS"] = string(SourceFile)
+	}
+	if src.LogLevel != "" && src.LogLevel != "INFO" {
+		dest.LogLevel = src.LogLevel
+		dest.sources["logLevel"] = string(SourceFile)
+	}
+	if src.InterfaceName != "" && src.InterfaceName != "olm" {
+		dest.InterfaceName = src.InterfaceName
+		dest.sources["interface"] = string(SourceFile)
+	}
+	if src.HTTPAddr != "" && src.HTTPAddr != ":9452" {
+		dest.HTTPAddr = src.HTTPAddr
+		dest.sources["httpAddr"] = string(SourceFile)
+	}
+	if src.SocketPath != "" {
+		// Check if it's not the default for any OS
+		isDefault := src.SocketPath == "/var/run/olm.sock" || src.SocketPath == "olm"
+		if !isDefault {
+			dest.SocketPath = src.SocketPath
+			dest.sources["socketPath"] = string(SourceFile)
+		}
+	}
+	if src.PingInterval != "" && src.PingInterval != "3s" {
+		dest.PingInterval = src.PingInterval
+		dest.sources["pingInterval"] = string(SourceFile)
+	}
+	if src.PingTimeout != "" && src.PingTimeout != "5s" {
+		dest.PingTimeout = src.PingTimeout
+		dest.sources["pingTimeout"] = string(SourceFile)
+	}
+	if src.TlsClientCert != "" {
+		dest.TlsClientCert = src.TlsClientCert
+		dest.sources["tlsClientCert"] = string(SourceFile)
+	}
+	// For booleans, we always take the source value if explicitly set
+	if src.EnableAPI {
+		dest.EnableAPI = src.EnableAPI
+		dest.sources["enableApi"] = string(SourceFile)
+	}
+	if src.DisableHolepunch {
+		dest.DisableHolepunch = src.DisableHolepunch
+		dest.sources["disableHolepunch"] = string(SourceFile)
+	}
+	if src.OverrideDNS {
+		dest.OverrideDNS = src.OverrideDNS
+		dest.sources["overrideDNS"] = string(SourceFile)
+	}
+	if src.DisableRelay {
+		dest.DisableRelay = src.DisableRelay
+		dest.sources["disableRelay"] = string(SourceFile)
+	}
+	// if src.DoNotCreateNewClient {
+	// 	dest.DoNotCreateNewClient = src.DoNotCreateNewClient
+	// 	dest.sources["doNotCreateNewClient"] = string(SourceFile)
+	// }
+}
+
+// SaveConfig saves the current configuration to the config file
+func SaveConfig(config *OlmConfig) error {
+	configPath := getOlmConfigPath()
+	data, err := json.MarshalIndent(config, "", "  ")
+	if err != nil {
+		return fmt.Errorf("failed to marshal config: %w", err)
+	}
+	return os.WriteFile(configPath, data, 0644)
+}
+
+// ShowConfig prints the configuration and the source of each value
+func (c *OlmConfig) ShowConfig() {
+	configPath := getOlmConfigPath()
+
+	fmt.Print("\n=== Olm Configuration ===\n\n")
+	fmt.Printf("Config File: %s\n", configPath)
+
+	// Check if config file exists
+	if _, err := os.Stat(configPath); err == nil {
+		fmt.Printf("Config File Status: ✓ exists\n")
+	} else {
+		fmt.Printf("Config File Status: ✗ not found\n")
+	}
+
+	fmt.Println("\n--- Configuration Values ---")
+	fmt.Print("(Format: Setting = Value [source])\n\n")
+
+	// Helper to get source or default
+	getSource := func(key string) string {
+		if source, ok := c.sources[key]; ok {
+			return source
+		}
+		return string(SourceDefault)
+	}
+
+	// Helper to format value (mask secrets)
+	formatValue := func(key, value string) string {
+		if key == "secret" && value != "" {
+			if len(value) > 8 {
+				return value[:4] + "****" + value[len(value)-4:]
+			}
+			return "****"
+		}
+		if value == "" {
+			return "(not set)"
+		}
+		return value
+	}
+
+	// Connection settings
+	fmt.Println("Connection:")
+	fmt.Printf("  endpoint     = %s [%s]\n", formatValue("endpoint", c.Endpoint), getSource("endpoint"))
+	fmt.Printf("  id           = %s [%s]\n", formatValue("id", c.ID), getSource("id"))
+	fmt.Printf("  secret       = %s [%s]\n", formatValue("secret", c.Secret), getSource("secret"))
+	fmt.Printf("  org          = %s [%s]\n", formatValue("org", c.OrgID), getSource("org"))
+	fmt.Printf("  user-token   = %s [%s]\n", formatValue("userToken", c.UserToken), getSource("userToken"))
+
+	// Network settings
+	fmt.Println("\nNetwork:")
+	fmt.Printf("  mtu          = %d [%s]\n", c.MTU, getSource("mtu"))
+	fmt.Printf("  dns          = %s [%s]\n", c.DNS, getSource("dns"))
+	fmt.Printf("  upstream-dns = %v [%s]\n", c.UpstreamDNS, getSource("upstreamDNS"))
+	fmt.Printf("  interface    = %s [%s]\n", c.InterfaceName, getSource("interface"))
+
+	// Logging
+	fmt.Println("\nLogging:")
+	fmt.Printf("  log-level    = %s [%s]\n", c.LogLevel, getSource("logLevel"))
+
+	// API server
+	fmt.Println("\nAPI Server:")
+	fmt.Printf("  enable-api   = %v [%s]\n", c.EnableAPI, getSource("enableApi"))
+	fmt.Printf("  http-addr    = %s [%s]\n", c.HTTPAddr, getSource("httpAddr"))
+	fmt.Printf("  socket-path  = %s [%s]\n", c.SocketPath, getSource("socketPath"))
+
+	// Timing
+	fmt.Println("\nTiming:")
+	fmt.Printf("  ping-interval = %s [%s]\n", c.PingInterval, getSource("pingInterval"))
+	fmt.Printf("  ping-timeout  = %s [%s]\n", c.PingTimeout, getSource("pingTimeout"))
+
+	// Advanced
+	fmt.Println("\nAdvanced:")
+	fmt.Printf("  disable-holepunch     = %v [%s]\n", c.DisableHolepunch, getSource("disableHolepunch"))
+	fmt.Printf("  override-dns          = %v [%s]\n", c.OverrideDNS, getSource("overrideDNS"))
+	fmt.Printf("  tunnel-dns            = %v [%s]\n", c.TunnelDNS, getSource("tunnelDNS"))
+	fmt.Printf("  disable-relay         = %v [%s]\n", c.DisableRelay, getSource("disableRelay"))
+	// fmt.Printf("  do-not-create-new-client = %v [%s]\n", c.DoNotCreateNewClient, getSource("doNotCreateNewClient"))
+	if c.TlsClientCert != "" {
+		fmt.Printf("  tls-cert              = %s [%s]\n", c.TlsClientCert, getSource("tlsClientCert"))
+	}
+
+	// Source legend
+	fmt.Println("\n--- Source Legend ---")
+	fmt.Println("  default     = Built-in default value")
+	fmt.Println("  file        = Loaded from config file")
+	fmt.Println("  environment = Set via environment variable")
+	fmt.Println("  cli         = Provided as command-line argument")
+	fmt.Println("\nPriority: cli > environment > file > default")
+	fmt.Println()
+}
+
+// splitComma splits a comma-separated string into a slice of trimmed strings
+func splitComma(s string) []string {
+	parts := strings.Split(s, ",")
+	result := make([]string, 0, len(parts))
+	for _, part := range parts {
+		trimmed := strings.TrimSpace(part)
+		if trimmed != "" {
+			result = append(result, trimmed)
+		}
+	}
+	return result
+}

create_test_creds.py

@@ -0,0 +1,43 @@
+
+import requests
+
+def create_olm(base_url, user_token, olm_name, user_id):
+    url = f"{base_url}/api/v1/user/{user_id}/olm"
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "User-Agent": "pangolin-cli",
+        "X-CSRF-Token": "x-csrf-protection",
+        "Cookie": f"p_session_token={user_token}"
+    }
+    payload = {"name": olm_name}
+    response = requests.put(url, json=payload, headers=headers)
+    response.raise_for_status()
+    data = response.json()
+    print(f"Response Data: {data}")
+
+def create_client(base_url, user_token, client_name):
+    url = f"{base_url}/api/v1/api/clients"
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "User-Agent": "pangolin-cli",
+        "X-CSRF-Token": "x-csrf-protection",
+        "Cookie": f"p_session_token={user_token}"
+    }
+    payload = {"name": client_name}
+    response = requests.post(url, json=payload, headers=headers)
+    response.raise_for_status()
+    data = response.json()
+    print(f"Response Data: {data}")
+
+if __name__ == "__main__":
+    # Example usage
+    base_url = input("Enter base URL (e.g., http://localhost:3000): ")
+    user_token = input("Enter user token: ")
+    user_id = input("Enter user ID: ")
+    olm_name = input("Enter OLM name: ")
+    client_name = input("Enter client name: ")
+
+    create_olm(base_url, user_token, olm_name, user_id)
+    # client_id = create_client(base_url, user_token, client_name)

device/middle_device.go

@@ -0,0 +1,331 @@
+package device
+
+import (
+	"net/netip"
+	"os"
+	"sync"
+
+	"github.com/fosrl/newt/logger"
+	"golang.zx2c4.com/wireguard/tun"
+)
+
+// PacketHandler processes intercepted packets and returns true if packet should be dropped
+type PacketHandler func(packet []byte) bool
+
+// FilterRule defines a rule for packet filtering
+type FilterRule struct {
+	DestIP  netip.Addr
+	Handler PacketHandler
+}
+
+// MiddleDevice wraps a TUN device with packet filtering capabilities
+type MiddleDevice struct {
+	tun.Device
+	rules    []FilterRule
+	mutex    sync.RWMutex
+	readCh   chan readResult
+	injectCh chan []byte
+	closed   chan struct{}
+}
+
+type readResult struct {
+	bufs   [][]byte
+	sizes  []int
+	offset int
+	n      int
+	err    error
+}
+
+// NewMiddleDevice creates a new filtered TUN device wrapper
+func NewMiddleDevice(device tun.Device) *MiddleDevice {
+	d := &MiddleDevice{
+		Device:   device,
+		rules:    make([]FilterRule, 0),
+		readCh:   make(chan readResult),
+		injectCh: make(chan []byte, 100),
+		closed:   make(chan struct{}),
+	}
+	go d.pump()
+	return d
+}
+
+func (d *MiddleDevice) pump() {
+	const defaultOffset = 16
+	batchSize := d.Device.BatchSize()
+	logger.Debug("MiddleDevice: pump started")
+
+	for {
+		// Check closed first with priority
+		select {
+		case <-d.closed:
+			logger.Debug("MiddleDevice: pump exiting due to closed channel")
+			return
+		default:
+		}
+
+		// Allocate buffers for reading
+		// We allocate new buffers for each read to avoid race conditions
+		// since we pass them to the channel
+		bufs := make([][]byte, batchSize)
+		sizes := make([]int, batchSize)
+		for i := range bufs {
+			bufs[i] = make([]byte, 2048) // Standard MTU + headroom
+		}
+
+		n, err := d.Device.Read(bufs, sizes, defaultOffset)
+
+		// Check closed again after read returns
+		select {
+		case <-d.closed:
+			logger.Debug("MiddleDevice: pump exiting due to closed channel (after read)")
+			return
+		default:
+		}
+
+		// Now try to send the result
+		select {
+		case d.readCh <- readResult{bufs: bufs, sizes: sizes, offset: defaultOffset, n: n, err: err}:
+		case <-d.closed:
+			logger.Debug("MiddleDevice: pump exiting due to closed channel (during send)")
+			return
+		}
+
+		if err != nil {
+			logger.Debug("MiddleDevice: pump exiting due to read error: %v", err)
+			return
+		}
+	}
+}
+
+// InjectOutbound injects a packet to be read by WireGuard (as if it came from TUN)
+func (d *MiddleDevice) InjectOutbound(packet []byte) {
+	select {
+	case d.injectCh <- packet:
+	case <-d.closed:
+	}
+}
+
+// AddRule adds a packet filtering rule
+func (d *MiddleDevice) AddRule(destIP netip.Addr, handler PacketHandler) {
+	d.mutex.Lock()
+	defer d.mutex.Unlock()
+	d.rules = append(d.rules, FilterRule{
+		DestIP:  destIP,
+		Handler: handler,
+	})
+}
+
+// RemoveRule removes all rules for a given destination IP
+func (d *MiddleDevice) RemoveRule(destIP netip.Addr) {
+	d.mutex.Lock()
+	defer d.mutex.Unlock()
+	newRules := make([]FilterRule, 0, len(d.rules))
+	for _, rule := range d.rules {
+		if rule.DestIP != destIP {
+			newRules = append(newRules, rule)
+		}
+	}
+	d.rules = newRules
+}
+
+// Close stops the device
+func (d *MiddleDevice) Close() error {
+	select {
+	case <-d.closed:
+		// Already closed
+		return nil
+	default:
+		logger.Debug("MiddleDevice: Closing, signaling closed channel")
+		close(d.closed)
+	}
+	logger.Debug("MiddleDevice: Closing underlying TUN device")
+	err := d.Device.Close()
+	logger.Debug("MiddleDevice: Underlying TUN device closed, err=%v", err)
+	return err
+}
+
+// extractDestIP extracts destination IP from packet (fast path)
+func extractDestIP(packet []byte) (netip.Addr, bool) {
+	if len(packet) < 20 {
+		return netip.Addr{}, false
+	}
+
+	version := packet[0] >> 4
+
+	switch version {
+	case 4:
+		if len(packet) < 20 {
+			return netip.Addr{}, false
+		}
+		// Destination IP is at bytes 16-19 for IPv4
+		ip := netip.AddrFrom4([4]byte{packet[16], packet[17], packet[18], packet[19]})
+		return ip, true
+	case 6:
+		if len(packet) < 40 {
+			return netip.Addr{}, false
+		}
+		// Destination IP is at bytes 24-39 for IPv6
+		var ip16 [16]byte
+		copy(ip16[:], packet[24:40])
+		ip := netip.AddrFrom16(ip16)
+		return ip, true
+	}
+
+	return netip.Addr{}, false
+}
+
+// Read intercepts packets going UP from the TUN device (towards WireGuard)
+func (d *MiddleDevice) Read(bufs [][]byte, sizes []int, offset int) (n int, err error) {
+	// Check if already closed first (non-blocking)
+	select {
+	case <-d.closed:
+		logger.Debug("MiddleDevice: Read returning os.ErrClosed (pre-check)")
+		return 0, os.ErrClosed
+	default:
+	}
+
+	// Now block waiting for data
+	select {
+	case res := <-d.readCh:
+		if res.err != nil {
+			logger.Debug("MiddleDevice: Read returning error from pump: %v", res.err)
+			return 0, res.err
+		}
+
+		// Copy packets from result to provided buffers
+		count := 0
+		for i := 0; i < res.n && i < len(bufs); i++ {
+			// Handle offset mismatch if necessary
+			// We assume the pump used defaultOffset (16)
+			// If caller asks for different offset, we need to shift
+			src := res.bufs[i]
+			srcOffset := res.offset
+			srcSize := res.sizes[i]
+
+			// Calculate where the packet data starts and ends in src
+			pktData := src[srcOffset : srcOffset+srcSize]
+
+			// Ensure dest buffer is large enough
+			if len(bufs[i]) < offset+len(pktData) {
+				continue // Skip if buffer too small
+			}
+
+			copy(bufs[i][offset:], pktData)
+			sizes[i] = len(pktData)
+			count++
+		}
+		n = count
+
+	case pkt := <-d.injectCh:
+		if len(bufs) == 0 {
+			return 0, nil
+		}
+		if len(bufs[0]) < offset+len(pkt) {
+			return 0, nil // Buffer too small
+		}
+		copy(bufs[0][offset:], pkt)
+		sizes[0] = len(pkt)
+		n = 1
+
+	case <-d.closed:
+		logger.Debug("MiddleDevice: Read returning os.ErrClosed")
+		return 0, os.ErrClosed // Signal that device is closed
+	}
+
+	d.mutex.RLock()
+	rules := d.rules
+	d.mutex.RUnlock()
+
+	if len(rules) == 0 {
+		return n, nil
+	}
+
+	// Process packets and filter out handled ones
+	writeIdx := 0
+	for readIdx := 0; readIdx < n; readIdx++ {
+		packet := bufs[readIdx][offset : offset+sizes[readIdx]]
+
+		destIP, ok := extractDestIP(packet)
+		if !ok {
+			// Can't parse, keep packet
+			if writeIdx != readIdx {
+				bufs[writeIdx] = bufs[readIdx]
+				sizes[writeIdx] = sizes[readIdx]
+			}
+			writeIdx++
+			continue
+		}
+
+		// Check if packet matches any rule
+		handled := false
+		for _, rule := range rules {
+			if rule.DestIP == destIP {
+				if rule.Handler(packet) {
+					// Packet was handled and should be dropped
+					handled = true
+					break
+				}
+			}
+		}
+
+		if !handled {
+			// Keep packet
+			if writeIdx != readIdx {
+				bufs[writeIdx] = bufs[readIdx]
+				sizes[writeIdx] = sizes[readIdx]
+			}
+			writeIdx++
+		}
+	}
+
+	return writeIdx, err
+}
+
+// Write intercepts packets going DOWN to the TUN device (from WireGuard)
+func (d *MiddleDevice) Write(bufs [][]byte, offset int) (int, error) {
+	d.mutex.RLock()
+	rules := d.rules
+	d.mutex.RUnlock()
+
+	if len(rules) == 0 {
+		return d.Device.Write(bufs, offset)
+	}
+
+	// Filter packets going down
+	filteredBufs := make([][]byte, 0, len(bufs))
+	for _, buf := range bufs {
+		if len(buf) <= offset {
+			continue
+		}
+
+		packet := buf[offset:]
+		destIP, ok := extractDestIP(packet)
+		if !ok {
+			// Can't parse, keep packet
+			filteredBufs = append(filteredBufs, buf)
+			continue
+		}
+
+		// Check if packet matches any rule
+		handled := false
+		for _, rule := range rules {
+			if rule.DestIP == destIP {
+				if rule.Handler(packet) {
+					// Packet was handled and should be dropped
+					handled = true
+					break
+				}
+			}
+		}
+
+		if !handled {
+			filteredBufs = append(filteredBufs, buf)
+		}
+	}
+
+	if len(filteredBufs) == 0 {
+		return len(bufs), nil // All packets were handled
+	}
+
+	return d.Device.Write(filteredBufs, offset)
+}

device/middle_device_test.go

@@ -0,0 +1,102 @@
+package device
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/fosrl/newt/util"
+)
+
+func TestExtractDestIP(t *testing.T) {
+	tests := []struct {
+		name   string
+		packet []byte
+		wantIP string
+		wantOk bool
+	}{
+		{
+			name: "IPv4 packet",
+			packet: []byte{
+				0x45, 0x00, 0x00, 0x54, 0x00, 0x00, 0x40, 0x00,
+				0x40, 0x11, 0x00, 0x00, 0xc0, 0xa8, 0x01, 0x01,
+				0x0a, 0x1e, 0x1e, 0x1e, // Dest IP: 10.30.30.30
+			},
+			wantIP: "10.30.30.30",
+			wantOk: true,
+		},
+		{
+			name:   "Too short packet",
+			packet: []byte{0x45, 0x00},
+			wantIP: "",
+			wantOk: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			gotIP, gotOk := extractDestIP(tt.packet)
+			if gotOk != tt.wantOk {
+				t.Errorf("extractDestIP() ok = %v, want %v", gotOk, tt.wantOk)
+				return
+			}
+			if tt.wantOk {
+				wantAddr := netip.MustParseAddr(tt.wantIP)
+				if gotIP != wantAddr {
+					t.Errorf("extractDestIP() ip = %v, want %v", gotIP, wantAddr)
+				}
+			}
+		})
+	}
+}
+
+func TestGetProtocol(t *testing.T) {
+	tests := []struct {
+		name      string
+		packet    []byte
+		wantProto uint8
+		wantOk    bool
+	}{
+		{
+			name: "UDP packet",
+			packet: []byte{
+				0x45, 0x00, 0x00, 0x54, 0x00, 0x00, 0x40, 0x00,
+				0x40, 0x11, 0x00, 0x00, 0xc0, 0xa8, 0x01, 0x01, // Protocol: UDP (17) at byte 9
+				0x0a, 0x1e, 0x1e, 0x1e,
+			},
+			wantProto: 17,
+			wantOk:    true,
+		},
+		{
+			name:      "Too short",
+			packet:    []byte{0x45, 0x00},
+			wantProto: 0,
+			wantOk:    false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			gotProto, gotOk := util.GetProtocol(tt.packet)
+			if gotOk != tt.wantOk {
+				t.Errorf("GetProtocol() ok = %v, want %v", gotOk, tt.wantOk)
+				return
+			}
+			if gotProto != tt.wantProto {
+				t.Errorf("GetProtocol() proto = %v, want %v", gotProto, tt.wantProto)
+			}
+		})
+	}
+}
+
+func BenchmarkExtractDestIP(b *testing.B) {
+	packet := []byte{
+		0x45, 0x00, 0x00, 0x54, 0x00, 0x00, 0x40, 0x00,
+		0x40, 0x11, 0x00, 0x00, 0xc0, 0xa8, 0x01, 0x01,
+		0x0a, 0x1e, 0x1e, 0x1e,
+	}
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		extractDestIP(packet)
+	}
+}

device/tun_unix.go

@@ -0,0 +1,44 @@
+//go:build !windows
+
+package device
+
+import (
+	"net"
+	"os"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/sys/unix"
+	"golang.zx2c4.com/wireguard/ipc"
+	"golang.zx2c4.com/wireguard/tun"
+)
+
+func CreateTUNFromFD(tunFd uint32, mtuInt int) (tun.Device, error) {
+	dupTunFd, err := unix.Dup(int(tunFd))
+	if err != nil {
+		logger.Error("Unable to dup tun fd: %v", err)
+		return nil, err
+	}
+
+	err = unix.SetNonblock(dupTunFd, true)
+	if err != nil {
+		unix.Close(dupTunFd)
+		return nil, err
+	}
+
+	file := os.NewFile(uintptr(dupTunFd), "/dev/tun")
+	device, err := tun.CreateTUNFromFile(file, mtuInt)
+	if err != nil {
+		file.Close()
+		return nil, err
+	}
+
+	return device, nil
+}
+
+func UapiOpen(interfaceName string) (*os.File, error) {
+	return ipc.UAPIOpen(interfaceName)
+}
+
+func UapiListen(interfaceName string, fileUAPI *os.File) (net.Listener, error) {
+	return ipc.UAPIListen(interfaceName, fileUAPI)
+}

device/tun_windows.go

@@ -0,0 +1,25 @@
+//go:build windows
+
+package device
+
+import (
+	"errors"
+	"net"
+	"os"
+
+	"golang.zx2c4.com/wireguard/ipc"
+	"golang.zx2c4.com/wireguard/tun"
+)
+
+func CreateTUNFromFD(tunFd uint32, mtuInt int) (tun.Device, error) {
+	return nil, errors.New("CreateTUNFromFile not supported on Windows")
+}
+
+func UapiOpen(interfaceName string) (*os.File, error) {
+	return nil, nil
+}
+
+func UapiListen(interfaceName string, fileUAPI *os.File) (net.Listener, error) {
+	// On Windows, UAPIListen only takes one parameter
+	return ipc.UAPIListen(interfaceName)
+}

dns/dns_proxy.go

@@ -0,0 +1,746 @@
+package dns
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"net/netip"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/util"
+	"github.com/fosrl/olm/device"
+	"github.com/miekg/dns"
+	"golang.zx2c4.com/wireguard/tun"
+	"gvisor.dev/gvisor/pkg/buffer"
+	"gvisor.dev/gvisor/pkg/tcpip"
+	"gvisor.dev/gvisor/pkg/tcpip/adapters/gonet"
+	"gvisor.dev/gvisor/pkg/tcpip/header"
+	"gvisor.dev/gvisor/pkg/tcpip/link/channel"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv6"
+	"gvisor.dev/gvisor/pkg/tcpip/stack"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/udp"
+)
+
+const (
+	DNSPort = 53
+)
+
+// DNSProxy implements a DNS proxy using gvisor netstack
+type DNSProxy struct {
+	stack        *stack.Stack
+	ep           *channel.Endpoint
+	proxyIP      netip.Addr
+	upstreamDNS  []string
+	tunnelDNS    bool                 // Whether to tunnel DNS queries over WireGuard or to spit them out locally
+	mtu          int
+	tunDevice    tun.Device           // Direct reference to underlying TUN device for responses
+	middleDevice *device.MiddleDevice // Reference to MiddleDevice for packet filtering
+	recordStore  *DNSRecordStore      // Local DNS records
+
+	// Tunnel DNS fields - for sending queries over WireGuard
+	tunnelIP         netip.Addr       // WireGuard interface IP (source for tunneled queries)
+	tunnelStack      *stack.Stack     // Separate netstack for outbound tunnel queries
+	tunnelEp         *channel.Endpoint
+	tunnelActivePorts map[uint16]bool
+	tunnelPortsLock  sync.Mutex
+
+	ctx    context.Context
+	cancel context.CancelFunc
+	wg     sync.WaitGroup
+}
+
+// NewDNSProxy creates a new DNS proxy
+func NewDNSProxy(tunDevice tun.Device, middleDevice *device.MiddleDevice, mtu int, utilitySubnet string, upstreamDns []string, tunnelDns bool, tunnelIP string) (*DNSProxy, error) {
+	proxyIP, err := PickIPFromSubnet(utilitySubnet)
+	if err != nil {
+		return nil, fmt.Errorf("failed to pick DNS proxy IP from subnet: %v", err)
+	}
+
+	if len(upstreamDns) == 0 {
+		return nil, fmt.Errorf("at least one upstream DNS server must be specified")
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	proxy := &DNSProxy{
+		proxyIP:           proxyIP,
+		mtu:               mtu,
+		tunDevice:         tunDevice,
+		middleDevice:      middleDevice,
+		upstreamDNS:       upstreamDns,
+		tunnelDNS:         tunnelDns,
+		recordStore:       NewDNSRecordStore(),
+		tunnelActivePorts: make(map[uint16]bool),
+		ctx:               ctx,
+		cancel:            cancel,
+	}
+
+	// Parse tunnel IP if provided (needed for tunneled DNS)
+	if tunnelIP != "" {
+		addr, err := netip.ParseAddr(tunnelIP)
+		if err != nil {
+			return nil, fmt.Errorf("failed to parse tunnel IP: %v", err)
+		}
+		proxy.tunnelIP = addr
+	}
+
+	// Create gvisor netstack for receiving DNS queries
+	stackOpts := stack.Options{
+		NetworkProtocols:   []stack.NetworkProtocolFactory{ipv4.NewProtocol, ipv6.NewProtocol},
+		TransportProtocols: []stack.TransportProtocolFactory{udp.NewProtocol},
+		HandleLocal:        true,
+	}
+
+	proxy.ep = channel.New(256, uint32(mtu), "")
+	proxy.stack = stack.New(stackOpts)
+
+	// Create NIC
+	if err := proxy.stack.CreateNIC(1, proxy.ep); err != nil {
+		return nil, fmt.Errorf("failed to create NIC: %v", err)
+	}
+
+	// Add IP address
+	// Parse the proxy IP to get the octets
+	ipBytes := proxyIP.As4()
+	protoAddr := tcpip.ProtocolAddress{
+		Protocol:          ipv4.ProtocolNumber,
+		AddressWithPrefix: tcpip.AddrFrom4(ipBytes).WithPrefix(),
+	}
+
+	if err := proxy.stack.AddProtocolAddress(1, protoAddr, stack.AddressProperties{}); err != nil {
+		return nil, fmt.Errorf("failed to add protocol address: %v", err)
+	}
+
+	// Add default route
+	proxy.stack.AddRoute(tcpip.Route{
+		Destination: header.IPv4EmptySubnet,
+		NIC:         1,
+	})
+
+	// Initialize tunnel netstack if tunnel DNS is enabled
+	if tunnelDns {
+		if !proxy.tunnelIP.IsValid() {
+			return nil, fmt.Errorf("tunnel IP is required when tunnelDNS is enabled")
+		}
+
+		// TODO: DO WE NEED TO ESTABLISH ANOTHER NETSTACK HERE OR CAN WE COMBINE WITH WGTESTER?
+		if err := proxy.initTunnelNetstack(); err != nil {
+			return nil, fmt.Errorf("failed to initialize tunnel netstack: %v", err)
+		}
+	}
+
+	return proxy, nil
+}
+
+// initTunnelNetstack creates a separate netstack for outbound DNS queries through the tunnel
+func (p *DNSProxy) initTunnelNetstack() error {
+	// Create gvisor netstack for outbound tunnel queries
+	stackOpts := stack.Options{
+		NetworkProtocols:   []stack.NetworkProtocolFactory{ipv4.NewProtocol, ipv6.NewProtocol},
+		TransportProtocols: []stack.TransportProtocolFactory{udp.NewProtocol},
+		HandleLocal:        true,
+	}
+
+	p.tunnelEp = channel.New(256, uint32(p.mtu), "")
+	p.tunnelStack = stack.New(stackOpts)
+
+	// Create NIC
+	if err := p.tunnelStack.CreateNIC(1, p.tunnelEp); err != nil {
+		return fmt.Errorf("failed to create tunnel NIC: %v", err)
+	}
+
+	// Add tunnel IP address (WireGuard interface IP)
+	ipBytes := p.tunnelIP.As4()
+	protoAddr := tcpip.ProtocolAddress{
+		Protocol:          ipv4.ProtocolNumber,
+		AddressWithPrefix: tcpip.AddrFrom4(ipBytes).WithPrefix(),
+	}
+
+	if err := p.tunnelStack.AddProtocolAddress(1, protoAddr, stack.AddressProperties{}); err != nil {
+		return fmt.Errorf("failed to add tunnel protocol address: %v", err)
+	}
+
+	// Add default route
+	p.tunnelStack.AddRoute(tcpip.Route{
+		Destination: header.IPv4EmptySubnet,
+		NIC:         1,
+	})
+
+	// Register filter rule on MiddleDevice to intercept responses
+	p.middleDevice.AddRule(p.tunnelIP, p.handleTunnelResponse)
+
+	return nil
+}
+
+// handleTunnelResponse handles packets coming back from the tunnel destined for the tunnel IP
+func (p *DNSProxy) handleTunnelResponse(packet []byte) bool {
+	// Check if it's UDP
+	proto, ok := util.GetProtocol(packet)
+	if !ok || proto != 17 { // UDP
+		return false
+	}
+
+	// Check destination port - should be one of our active outbound ports
+	port, ok := util.GetDestPort(packet)
+	if !ok {
+		return false
+	}
+
+	// Check if we are expecting a response on this port
+	p.tunnelPortsLock.Lock()
+	active := p.tunnelActivePorts[uint16(port)]
+	p.tunnelPortsLock.Unlock()
+
+	if !active {
+		return false
+	}
+
+	// Inject into tunnel netstack
+	version := packet[0] >> 4
+	pkb := stack.NewPacketBuffer(stack.PacketBufferOptions{
+		Payload: buffer.MakeWithData(packet),
+	})
+
+	switch version {
+	case 4:
+		p.tunnelEp.InjectInbound(ipv4.ProtocolNumber, pkb)
+	case 6:
+		p.tunnelEp.InjectInbound(ipv6.ProtocolNumber, pkb)
+	default:
+		pkb.DecRef()
+		return false
+	}
+
+	pkb.DecRef()
+	return true // Handled
+}
+
+// Start starts the DNS proxy and registers with the filter
+func (p *DNSProxy) Start() error {
+	// Install packet filter rule
+	p.middleDevice.AddRule(p.proxyIP, p.handlePacket)
+
+	// Start DNS listener
+	p.wg.Add(2)
+	go p.runDNSListener()
+	go p.runPacketSender()
+
+	// Start tunnel packet sender if tunnel DNS is enabled
+	if p.tunnelDNS {
+		p.wg.Add(1)
+		go p.runTunnelPacketSender()
+	}
+
+	logger.Info("DNS proxy started on %s:%d (tunnelDNS=%v)", p.proxyIP.String(), DNSPort, p.tunnelDNS)
+	return nil
+}
+
+// Stop stops the DNS proxy
+func (p *DNSProxy) Stop() {
+	if p.middleDevice != nil {
+		p.middleDevice.RemoveRule(p.proxyIP)
+		if p.tunnelDNS && p.tunnelIP.IsValid() {
+			p.middleDevice.RemoveRule(p.tunnelIP)
+		}
+	}
+	p.cancel()
+
+	// Close the endpoint first to unblock any pending Read() calls in runPacketSender
+	if p.ep != nil {
+		p.ep.Close()
+	}
+
+	// Close tunnel endpoint if it exists
+	if p.tunnelEp != nil {
+		p.tunnelEp.Close()
+	}
+
+	p.wg.Wait()
+
+	if p.stack != nil {
+		p.stack.Close()
+	}
+
+	if p.tunnelStack != nil {
+		p.tunnelStack.Close()
+	}
+
+	logger.Info("DNS proxy stopped")
+}
+
+func (p *DNSProxy) GetProxyIP() netip.Addr {
+	return p.proxyIP
+}
+
+// handlePacket is called by the filter for packets destined to DNS proxy IP
+func (p *DNSProxy) handlePacket(packet []byte) bool {
+	if len(packet) < 20 {
+		return false // Don't drop, malformed
+	}
+
+	// Quick check for UDP port 53
+	proto, ok := util.GetProtocol(packet)
+	if !ok || proto != 17 { // 17 = UDP
+		return false // Not UDP, don't handle
+	}
+
+	port, ok := util.GetDestPort(packet)
+	if !ok || port != DNSPort {
+		return false // Not DNS port
+	}
+
+	// Inject packet into our netstack
+	version := packet[0] >> 4
+	pkb := stack.NewPacketBuffer(stack.PacketBufferOptions{
+		Payload: buffer.MakeWithData(packet),
+	})
+
+	switch version {
+	case 4:
+		p.ep.InjectInbound(ipv4.ProtocolNumber, pkb)
+	case 6:
+		p.ep.InjectInbound(ipv6.ProtocolNumber, pkb)
+	default:
+		pkb.DecRef()
+		return false
+	}
+
+	pkb.DecRef()
+	return true // Drop packet from normal path
+}
+
+// runDNSListener listens for DNS queries on the netstack
+func (p *DNSProxy) runDNSListener() {
+	defer p.wg.Done()
+
+	// Create UDP listener using gonet
+	// Parse the proxy IP to get the octets
+	ipBytes := p.proxyIP.As4()
+	laddr := &tcpip.FullAddress{
+		NIC:  1,
+		Addr: tcpip.AddrFrom4(ipBytes),
+		Port: DNSPort,
+	}
+
+	udpConn, err := gonet.DialUDP(p.stack, laddr, nil, ipv4.ProtocolNumber)
+	if err != nil {
+		logger.Error("Failed to create DNS listener: %v", err)
+		return
+	}
+	defer udpConn.Close()
+
+	logger.Debug("DNS proxy listening on netstack")
+
+	// Handle DNS queries
+	buf := make([]byte, 4096)
+	for {
+		select {
+		case <-p.ctx.Done():
+			return
+		default:
+		}
+
+		udpConn.SetReadDeadline(time.Now().Add(1 * time.Second))
+		n, remoteAddr, err := udpConn.ReadFrom(buf)
+		if err != nil {
+			if netErr, ok := err.(net.Error); ok && netErr.Timeout() {
+				continue
+			}
+			if p.ctx.Err() != nil {
+				return
+			}
+			logger.Error("DNS read error: %v", err)
+			continue
+		}
+
+		query := make([]byte, n)
+		copy(query, buf[:n])
+
+		// Handle query in background
+		go p.handleDNSQuery(udpConn, query, remoteAddr)
+	}
+}
+
+// handleDNSQuery processes a DNS query, checking local records first, then forwarding upstream
+func (p *DNSProxy) handleDNSQuery(udpConn *gonet.UDPConn, queryData []byte, clientAddr net.Addr) {
+	// Parse the DNS query
+	msg := new(dns.Msg)
+	if err := msg.Unpack(queryData); err != nil {
+		logger.Error("Failed to parse DNS query: %v", err)
+		return
+	}
+
+	if len(msg.Question) == 0 {
+		logger.Debug("DNS query has no questions")
+		return
+	}
+
+	question := msg.Question[0]
+	logger.Debug("DNS query for %s (type %s)", question.Name, dns.TypeToString[question.Qtype])
+
+	// Check if we have local records for this query
+	var response *dns.Msg
+	if question.Qtype == dns.TypeA || question.Qtype == dns.TypeAAAA {
+		response = p.checkLocalRecords(msg, question)
+	}
+
+	// If no local records, forward to upstream
+	if response == nil {
+		logger.Debug("No local record for %s, forwarding upstream to %v", question.Name, p.upstreamDNS)
+		response = p.forwardToUpstream(msg)
+	}
+
+	if response == nil {
+		logger.Error("Failed to get DNS response for %s", question.Name)
+		return
+	}
+
+	// Pack and send response
+	responseData, err := response.Pack()
+	if err != nil {
+		logger.Error("Failed to pack DNS response: %v", err)
+		return
+	}
+
+	_, err = udpConn.WriteTo(responseData, clientAddr)
+	if err != nil {
+		logger.Error("Failed to send DNS response: %v", err)
+	}
+}
+
+// checkLocalRecords checks if we have local records for the query
+func (p *DNSProxy) checkLocalRecords(query *dns.Msg, question dns.Question) *dns.Msg {
+	var recordType RecordType
+	if question.Qtype == dns.TypeA {
+		recordType = RecordTypeA
+	} else if question.Qtype == dns.TypeAAAA {
+		recordType = RecordTypeAAAA
+	} else {
+		return nil
+	}
+
+	ips := p.recordStore.GetRecords(question.Name, recordType)
+	if len(ips) == 0 {
+		return nil
+	}
+
+	logger.Debug("Found %d local record(s) for %s", len(ips), question.Name)
+
+	// Create response message
+	response := new(dns.Msg)
+	response.SetReply(query)
+	response.Authoritative = true
+
+	// Add answer records
+	for _, ip := range ips {
+		var rr dns.RR
+		if question.Qtype == dns.TypeA {
+			rr = &dns.A{
+				Hdr: dns.RR_Header{
+					Name:   question.Name,
+					Rrtype: dns.TypeA,
+					Class:  dns.ClassINET,
+					Ttl:    300, // 5 minutes
+				},
+				A: ip.To4(),
+			}
+		} else { // TypeAAAA
+			rr = &dns.AAAA{
+				Hdr: dns.RR_Header{
+					Name:   question.Name,
+					Rrtype: dns.TypeAAAA,
+					Class:  dns.ClassINET,
+					Ttl:    300, // 5 minutes
+				},
+				AAAA: ip.To16(),
+			}
+		}
+		response.Answer = append(response.Answer, rr)
+	}
+
+	return response
+}
+
+// forwardToUpstream forwards a DNS query to upstream DNS servers
+func (p *DNSProxy) forwardToUpstream(query *dns.Msg) *dns.Msg {
+	// Try primary DNS server
+	response, err := p.queryUpstream(p.upstreamDNS[0], query, 2*time.Second)
+	if err != nil && len(p.upstreamDNS) > 1 {
+		// Try secondary DNS server
+		logger.Debug("Primary DNS failed, trying secondary: %v", err)
+		response, err = p.queryUpstream(p.upstreamDNS[1], query, 2*time.Second)
+		if err != nil {
+			logger.Error("Both DNS servers failed: %v", err)
+			return nil
+		}
+	}
+	return response
+}
+
+// queryUpstream sends a DNS query to upstream server
+func (p *DNSProxy) queryUpstream(server string, query *dns.Msg, timeout time.Duration) (*dns.Msg, error) {
+	if p.tunnelDNS {
+		return p.queryUpstreamTunnel(server, query, timeout)
+	}
+	return p.queryUpstreamDirect(server, query, timeout)
+}
+
+// queryUpstreamDirect sends a DNS query to upstream server using miekg/dns directly (host networking)
+func (p *DNSProxy) queryUpstreamDirect(server string, query *dns.Msg, timeout time.Duration) (*dns.Msg, error) {
+	client := &dns.Client{
+		Timeout: timeout,
+	}
+
+	response, _, err := client.Exchange(query, server)
+	if err != nil {
+		return nil, err
+	}
+
+	return response, nil
+}
+
+// queryUpstreamTunnel sends a DNS query through the WireGuard tunnel
+func (p *DNSProxy) queryUpstreamTunnel(server string, query *dns.Msg, timeout time.Duration) (*dns.Msg, error) {
+	// Dial through the tunnel netstack
+	conn, port, err := p.dialTunnel("udp", server)
+	if err != nil {
+		return nil, fmt.Errorf("failed to dial tunnel: %v", err)
+	}
+	defer func() {
+		conn.Close()
+		p.removeTunnelPort(port)
+	}()
+
+	// Pack the query
+	queryData, err := query.Pack()
+	if err != nil {
+		return nil, fmt.Errorf("failed to pack query: %v", err)
+	}
+
+	// Set deadline
+	conn.SetDeadline(time.Now().Add(timeout))
+
+	// Send the query
+	_, err = conn.Write(queryData)
+	if err != nil {
+		return nil, fmt.Errorf("failed to send query: %v", err)
+	}
+
+	// Read the response
+	buf := make([]byte, 4096)
+	n, err := conn.Read(buf)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read response: %v", err)
+	}
+
+	// Parse the response
+	response := new(dns.Msg)
+	if err := response.Unpack(buf[:n]); err != nil {
+		return nil, fmt.Errorf("failed to unpack response: %v", err)
+	}
+
+	return response, nil
+}
+
+// dialTunnel creates a UDP connection through the tunnel netstack
+func (p *DNSProxy) dialTunnel(network, addr string) (net.Conn, uint16, error) {
+	if p.tunnelStack == nil {
+		return nil, 0, fmt.Errorf("tunnel netstack not initialized")
+	}
+
+	// Parse remote address
+	raddr, err := net.ResolveUDPAddr("udp", addr)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	// Use tunnel IP as source
+	ipBytes := p.tunnelIP.As4()
+
+	// Create UDP connection with ephemeral port
+	laddr := &tcpip.FullAddress{
+		NIC:  1,
+		Addr: tcpip.AddrFrom4(ipBytes),
+		Port: 0,
+	}
+
+	raddrTcpip := &tcpip.FullAddress{
+		NIC:  1,
+		Addr: tcpip.AddrFrom4([4]byte(raddr.IP.To4())),
+		Port: uint16(raddr.Port),
+	}
+
+	conn, err := gonet.DialUDP(p.tunnelStack, laddr, raddrTcpip, ipv4.ProtocolNumber)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	// Get local port
+	localAddr := conn.LocalAddr().(*net.UDPAddr)
+	port := uint16(localAddr.Port)
+
+	// Register port so we can receive responses
+	p.tunnelPortsLock.Lock()
+	p.tunnelActivePorts[port] = true
+	p.tunnelPortsLock.Unlock()
+
+	return conn, port, nil
+}
+
+// removeTunnelPort removes a port from the active ports map
+func (p *DNSProxy) removeTunnelPort(port uint16) {
+	p.tunnelPortsLock.Lock()
+	delete(p.tunnelActivePorts, port)
+	p.tunnelPortsLock.Unlock()
+}
+
+// runTunnelPacketSender reads packets from tunnel netstack and injects them into WireGuard
+func (p *DNSProxy) runTunnelPacketSender() {
+	defer p.wg.Done()
+	logger.Debug("DNS tunnel packet sender goroutine started")
+
+	ticker := time.NewTicker(1 * time.Millisecond)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-p.ctx.Done():
+			logger.Debug("DNS tunnel packet sender exiting")
+			// Drain any remaining packets
+			for {
+				pkt := p.tunnelEp.Read()
+				if pkt == nil {
+					break
+				}
+				pkt.DecRef()
+			}
+			return
+		case <-ticker.C:
+			// Try to read packets
+			for i := 0; i < 10; i++ {
+				pkt := p.tunnelEp.Read()
+				if pkt == nil {
+					break
+				}
+
+				// Extract packet data
+				slices := pkt.AsSlices()
+				if len(slices) > 0 {
+					var totalSize int
+					for _, slice := range slices {
+						totalSize += len(slice)
+					}
+
+					buf := make([]byte, totalSize)
+					pos := 0
+					for _, slice := range slices {
+						copy(buf[pos:], slice)
+						pos += len(slice)
+					}
+
+					// Inject into MiddleDevice (outbound to WG)
+					p.middleDevice.InjectOutbound(buf)
+				}
+
+				pkt.DecRef()
+			}
+		}
+	}
+}
+
+// runPacketSender sends packets from netstack back to TUN
+func (p *DNSProxy) runPacketSender() {
+	defer p.wg.Done()
+
+	// MessageTransportHeaderSize is the offset used by WireGuard device
+	// for reading/writing packets to the TUN interface
+	const offset = 16
+
+	for {
+		select {
+		case <-p.ctx.Done():
+			return
+		default:
+		}
+
+		// Read packets from netstack endpoint
+		pkt := p.ep.Read()
+		if pkt == nil {
+			// No packet available, small sleep to avoid busy loop
+			time.Sleep(1 * time.Millisecond)
+			continue
+		}
+
+		// Extract packet data as slices
+		slices := pkt.AsSlices()
+		if len(slices) > 0 {
+			// Flatten all slices into a single packet buffer
+			var totalSize int
+			for _, slice := range slices {
+				totalSize += len(slice)
+			}
+
+			// Allocate buffer with offset space for WireGuard transport header
+			// The first 'offset' bytes are reserved for the transport header
+			buf := make([]byte, offset+totalSize)
+
+			// Copy packet data after the offset
+			pos := offset
+			for _, slice := range slices {
+				copy(buf[pos:], slice)
+				pos += len(slice)
+			}
+
+			// Write packet to TUN device
+			// offset=16 indicates packet data starts at position 16 in the buffer
+			_, err := p.tunDevice.Write([][]byte{buf}, offset)
+			if err != nil {
+				logger.Error("Failed to write DNS response to TUN: %v", err)
+			}
+		}
+
+		pkt.DecRef()
+	}
+}
+
+// AddDNSRecord adds a DNS record to the local store
+// domain should be a domain name (e.g., "example.com" or "example.com.")
+// ip should be a valid IPv4 or IPv6 address
+func (p *DNSProxy) AddDNSRecord(domain string, ip net.IP) error {
+	return p.recordStore.AddRecord(domain, ip)
+}
+
+// RemoveDNSRecord removes a DNS record from the local store
+// If ip is nil, removes all records for the domain
+func (p *DNSProxy) RemoveDNSRecord(domain string, ip net.IP) {
+	p.recordStore.RemoveRecord(domain, ip)
+}
+
+// GetDNSRecords returns all IP addresses for a domain and record type
+func (p *DNSProxy) GetDNSRecords(domain string, recordType RecordType) []net.IP {
+	return p.recordStore.GetRecords(domain, recordType)
+}
+
+// ClearDNSRecords removes all DNS records from the local store
+func (p *DNSProxy) ClearDNSRecords() {
+	p.recordStore.Clear()
+}
+
+func PickIPFromSubnet(subnet string) (netip.Addr, error) {
+	// given a subnet in CIDR notation, pick the first usable IP
+	prefix, err := netip.ParsePrefix(subnet)
+	if err != nil {
+		return netip.Addr{}, fmt.Errorf("invalid subnet: %w", err)
+	}
+
+	// Pick the first usable IP address from the subnet
+	ip := prefix.Addr().Next()
+	if !ip.IsValid() {
+		return netip.Addr{}, fmt.Errorf("no valid IP address found in subnet: %s", subnet)
+	}
+
+	return ip, nil
+}

dns/dns_records.go

@@ -0,0 +1,325 @@
+package dns
+
+import (
+	"net"
+	"strings"
+	"sync"
+
+	"github.com/miekg/dns"
+)
+
+// RecordType represents the type of DNS record
+type RecordType uint16
+
+const (
+	RecordTypeA    RecordType = RecordType(dns.TypeA)
+	RecordTypeAAAA RecordType = RecordType(dns.TypeAAAA)
+)
+
+// DNSRecordStore manages local DNS records for A and AAAA queries
+type DNSRecordStore struct {
+	mu            sync.RWMutex
+	aRecords      map[string][]net.IP // domain -> list of IPv4 addresses
+	aaaaRecords   map[string][]net.IP // domain -> list of IPv6 addresses
+	aWildcards    map[string][]net.IP // wildcard pattern -> list of IPv4 addresses
+	aaaaWildcards map[string][]net.IP // wildcard pattern -> list of IPv6 addresses
+}
+
+// NewDNSRecordStore creates a new DNS record store
+func NewDNSRecordStore() *DNSRecordStore {
+	return &DNSRecordStore{
+		aRecords:      make(map[string][]net.IP),
+		aaaaRecords:   make(map[string][]net.IP),
+		aWildcards:    make(map[string][]net.IP),
+		aaaaWildcards: make(map[string][]net.IP),
+	}
+}
+
+// AddRecord adds a DNS record mapping (A or AAAA)
+// domain should be in FQDN format (e.g., "example.com.")
+// domain can contain wildcards: * (0+ chars) and ? (exactly 1 char)
+// ip should be a valid IPv4 or IPv6 address
+func (s *DNSRecordStore) AddRecord(domain string, ip net.IP) error {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
+	// Ensure domain ends with a dot (FQDN format)
+	if len(domain) == 0 || domain[len(domain)-1] != '.' {
+		domain = domain + "."
+	}
+
+	// Normalize domain to lowercase
+	domain = dns.Fqdn(domain)
+
+	// Check if domain contains wildcards
+	isWildcard := strings.ContainsAny(domain, "*?")
+
+	if ip.To4() != nil {
+		// IPv4 address
+		if isWildcard {
+			s.aWildcards[domain] = append(s.aWildcards[domain], ip)
+		} else {
+			s.aRecords[domain] = append(s.aRecords[domain], ip)
+		}
+	} else if ip.To16() != nil {
+		// IPv6 address
+		if isWildcard {
+			s.aaaaWildcards[domain] = append(s.aaaaWildcards[domain], ip)
+		} else {
+			s.aaaaRecords[domain] = append(s.aaaaRecords[domain], ip)
+		}
+	} else {
+		return &net.ParseError{Type: "IP address", Text: ip.String()}
+	}
+
+	return nil
+}
+
+// RemoveRecord removes a specific DNS record mapping
+// If ip is nil, removes all records for the domain (including wildcards)
+func (s *DNSRecordStore) RemoveRecord(domain string, ip net.IP) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
+	// Ensure domain ends with a dot (FQDN format)
+	if len(domain) == 0 || domain[len(domain)-1] != '.' {
+		domain = domain + "."
+	}
+
+	// Normalize domain to lowercase
+	domain = dns.Fqdn(domain)
+
+	// Check if domain contains wildcards
+	isWildcard := strings.ContainsAny(domain, "*?")
+
+	if ip == nil {
+		// Remove all records for this domain
+		if isWildcard {
+			delete(s.aWildcards, domain)
+			delete(s.aaaaWildcards, domain)
+		} else {
+			delete(s.aRecords, domain)
+			delete(s.aaaaRecords, domain)
+		}
+		return
+	}
+
+	if ip.To4() != nil {
+		// Remove specific IPv4 address
+		if isWildcard {
+			if ips, ok := s.aWildcards[domain]; ok {
+				s.aWildcards[domain] = removeIP(ips, ip)
+				if len(s.aWildcards[domain]) == 0 {
+					delete(s.aWildcards, domain)
+				}
+			}
+		} else {
+			if ips, ok := s.aRecords[domain]; ok {
+				s.aRecords[domain] = removeIP(ips, ip)
+				if len(s.aRecords[domain]) == 0 {
+					delete(s.aRecords, domain)
+				}
+			}
+		}
+	} else if ip.To16() != nil {
+		// Remove specific IPv6 address
+		if isWildcard {
+			if ips, ok := s.aaaaWildcards[domain]; ok {
+				s.aaaaWildcards[domain] = removeIP(ips, ip)
+				if len(s.aaaaWildcards[domain]) == 0 {
+					delete(s.aaaaWildcards, domain)
+				}
+			}
+		} else {
+			if ips, ok := s.aaaaRecords[domain]; ok {
+				s.aaaaRecords[domain] = removeIP(ips, ip)
+				if len(s.aaaaRecords[domain]) == 0 {
+					delete(s.aaaaRecords, domain)
+				}
+			}
+		}
+	}
+}
+
+// GetRecords returns all IP addresses for a domain and record type
+// First checks for exact matches, then checks wildcard patterns
+func (s *DNSRecordStore) GetRecords(domain string, recordType RecordType) []net.IP {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+
+	// Normalize domain to lowercase FQDN
+	domain = dns.Fqdn(domain)
+
+	var records []net.IP
+	switch recordType {
+	case RecordTypeA:
+		// Check exact match first
+		if ips, ok := s.aRecords[domain]; ok {
+			// Return a copy to prevent external modifications
+			records = make([]net.IP, len(ips))
+			copy(records, ips)
+			return records
+		}
+		// Check wildcard patterns
+		for pattern, ips := range s.aWildcards {
+			if matchWildcard(pattern, domain) {
+				records = append(records, ips...)
+			}
+		}
+		if len(records) > 0 {
+			// Return a copy
+			result := make([]net.IP, len(records))
+			copy(result, records)
+			return result
+		}
+
+	case RecordTypeAAAA:
+		// Check exact match first
+		if ips, ok := s.aaaaRecords[domain]; ok {
+			// Return a copy to prevent external modifications
+			records = make([]net.IP, len(ips))
+			copy(records, ips)
+			return records
+		}
+		// Check wildcard patterns
+		for pattern, ips := range s.aaaaWildcards {
+			if matchWildcard(pattern, domain) {
+				records = append(records, ips...)
+			}
+		}
+		if len(records) > 0 {
+			// Return a copy
+			result := make([]net.IP, len(records))
+			copy(result, records)
+			return result
+		}
+	}
+
+	return records
+}
+
+// HasRecord checks if a domain has any records of the specified type
+// Checks both exact matches and wildcard patterns
+func (s *DNSRecordStore) HasRecord(domain string, recordType RecordType) bool {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+
+	// Normalize domain to lowercase FQDN
+	domain = dns.Fqdn(domain)
+
+	switch recordType {
+	case RecordTypeA:
+		// Check exact match
+		if _, ok := s.aRecords[domain]; ok {
+			return true
+		}
+		// Check wildcard patterns
+		for pattern := range s.aWildcards {
+			if matchWildcard(pattern, domain) {
+				return true
+			}
+		}
+	case RecordTypeAAAA:
+		// Check exact match
+		if _, ok := s.aaaaRecords[domain]; ok {
+			return true
+		}
+		// Check wildcard patterns
+		for pattern := range s.aaaaWildcards {
+			if matchWildcard(pattern, domain) {
+				return true
+			}
+		}
+	}
+
+	return false
+}
+
+// Clear removes all records from the store
+func (s *DNSRecordStore) Clear() {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
+	s.aRecords = make(map[string][]net.IP)
+	s.aaaaRecords = make(map[string][]net.IP)
+	s.aWildcards = make(map[string][]net.IP)
+	s.aaaaWildcards = make(map[string][]net.IP)
+}
+
+// removeIP is a helper function to remove a specific IP from a slice
+func removeIP(ips []net.IP, toRemove net.IP) []net.IP {
+	result := make([]net.IP, 0, len(ips))
+	for _, ip := range ips {
+		if !ip.Equal(toRemove) {
+			result = append(result, ip)
+		}
+	}
+	return result
+}
+
+// matchWildcard checks if a domain matches a wildcard pattern
+// Pattern supports * (0+ chars) and ? (exactly 1 char)
+// Special case: *.domain.com does not match domain.com itself
+func matchWildcard(pattern, domain string) bool {
+	return matchWildcardInternal(pattern, domain, 0, 0)
+}
+
+// matchWildcardInternal performs the actual wildcard matching recursively
+func matchWildcardInternal(pattern, domain string, pi, di int) bool {
+	plen := len(pattern)
+	dlen := len(domain)
+
+	// Base cases
+	if pi == plen && di == dlen {
+		return true
+	}
+	if pi == plen {
+		return false
+	}
+
+	// Handle wildcard characters
+	if pattern[pi] == '*' {
+		// Special case: if pattern starts with "*." and we're at the beginning,
+		// ensure we don't match the domain without a prefix
+		// e.g., *.autoco.internal should not match autoco.internal
+		if pi == 0 && pi+1 < plen && pattern[pi+1] == '.' {
+			// The * must match at least one character
+			if di == dlen {
+				return false
+			}
+			// Try matching 1 or more characters before the dot
+			for i := di + 1; i <= dlen; i++ {
+				if matchWildcardInternal(pattern, domain, pi+1, i) {
+					return true
+				}
+			}
+			return false
+		}
+
+		// Normal * matching (0 or more characters)
+		// Try matching 0 characters (skip the *)
+		if matchWildcardInternal(pattern, domain, pi+1, di) {
+			return true
+		}
+		// Try matching 1+ characters
+		if di < dlen {
+			return matchWildcardInternal(pattern, domain, pi, di+1)
+		}
+		return false
+	}
+
+	if pattern[pi] == '?' {
+		// ? matches exactly one character
+		if di >= dlen {
+			return false
+		}
+		return matchWildcardInternal(pattern, domain, pi+1, di+1)
+	}
+
+	// Regular character - must match exactly
+	if di >= dlen || pattern[pi] != domain[di] {
+		return false
+	}
+
+	return matchWildcardInternal(pattern, domain, pi+1, di+1)
+}

dns/dns_records_test.go

@@ -0,0 +1,350 @@
+package dns
+
+import (
+	"net"
+	"testing"
+)
+
+func TestWildcardMatching(t *testing.T) {
+	tests := []struct {
+		name     string
+		pattern  string
+		domain   string
+		expected bool
+	}{
+		// Basic wildcard tests
+		{
+			name:     "*.autoco.internal matches host.autoco.internal",
+			pattern:  "*.autoco.internal.",
+			domain:   "host.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "*.autoco.internal matches longerhost.autoco.internal",
+			pattern:  "*.autoco.internal.",
+			domain:   "longerhost.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "*.autoco.internal matches sub.host.autoco.internal",
+			pattern:  "*.autoco.internal.",
+			domain:   "sub.host.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "*.autoco.internal does NOT match autoco.internal",
+			pattern:  "*.autoco.internal.",
+			domain:   "autoco.internal.",
+			expected: false,
+		},
+		
+		// Question mark wildcard tests
+		{
+			name:     "host-0?.autoco.internal matches host-01.autoco.internal",
+			pattern:  "host-0?.autoco.internal.",
+			domain:   "host-01.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "host-0?.autoco.internal matches host-0a.autoco.internal",
+			pattern:  "host-0?.autoco.internal.",
+			domain:   "host-0a.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "host-0?.autoco.internal does NOT match host-0.autoco.internal",
+			pattern:  "host-0?.autoco.internal.",
+			domain:   "host-0.autoco.internal.",
+			expected: false,
+		},
+		{
+			name:     "host-0?.autoco.internal does NOT match host-012.autoco.internal",
+			pattern:  "host-0?.autoco.internal.",
+			domain:   "host-012.autoco.internal.",
+			expected: false,
+		},
+		
+		// Combined wildcard tests
+		{
+			name:     "*.host-0?.autoco.internal matches sub.host-01.autoco.internal",
+			pattern:  "*.host-0?.autoco.internal.",
+			domain:   "sub.host-01.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "*.host-0?.autoco.internal matches prefix.host-0a.autoco.internal",
+			pattern:  "*.host-0?.autoco.internal.",
+			domain:   "prefix.host-0a.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "*.host-0?.autoco.internal does NOT match host-01.autoco.internal",
+			pattern:  "*.host-0?.autoco.internal.",
+			domain:   "host-01.autoco.internal.",
+			expected: false,
+		},
+		
+		// Multiple asterisks
+		{
+			name:     "*.*. autoco.internal matches any.thing.autoco.internal",
+			pattern:  "*.*.autoco.internal.",
+			domain:   "any.thing.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "*.*.autoco.internal does NOT match single.autoco.internal",
+			pattern:  "*.*.autoco.internal.",
+			domain:   "single.autoco.internal.",
+			expected: false,
+		},
+		
+		// Asterisk in middle
+		{
+			name:     "host-*.autoco.internal matches host-anything.autoco.internal",
+			pattern:  "host-*.autoco.internal.",
+			domain:   "host-anything.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "host-*.autoco.internal matches host-.autoco.internal (empty match)",
+			pattern:  "host-*.autoco.internal.",
+			domain:   "host-.autoco.internal.",
+			expected: true,
+		},
+		
+		// Multiple question marks
+		{
+			name:     "host-??.autoco.internal matches host-01.autoco.internal",
+			pattern:  "host-??.autoco.internal.",
+			domain:   "host-01.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "host-??.autoco.internal does NOT match host-1.autoco.internal",
+			pattern:  "host-??.autoco.internal.",
+			domain:   "host-1.autoco.internal.",
+			expected: false,
+		},
+		
+		// Exact match (no wildcards)
+		{
+			name:     "exact.autoco.internal matches exact.autoco.internal",
+			pattern:  "exact.autoco.internal.",
+			domain:   "exact.autoco.internal.",
+			expected: true,
+		},
+		{
+			name:     "exact.autoco.internal does NOT match other.autoco.internal",
+			pattern:  "exact.autoco.internal.",
+			domain:   "other.autoco.internal.",
+			expected: false,
+		},
+		
+		// Edge cases
+		{
+			name:     "* matches anything",
+			pattern:  "*",
+			domain:   "anything.at.all.",
+			expected: true,
+		},
+		{
+			name:     "*.* matches multi.level.",
+			pattern:  "*.*",
+			domain:   "multi.level.",
+			expected: true,
+		},
+	}
+	
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := matchWildcard(tt.pattern, tt.domain)
+			if result != tt.expected {
+				t.Errorf("matchWildcard(%q, %q) = %v, want %v", tt.pattern, tt.domain, result, tt.expected)
+			}
+		})
+	}
+}
+
+func TestDNSRecordStoreWildcard(t *testing.T) {
+	store := NewDNSRecordStore()
+	
+	// Add wildcard records
+	wildcardIP := net.ParseIP("10.0.0.1")
+	err := store.AddRecord("*.autoco.internal", wildcardIP)
+	if err != nil {
+		t.Fatalf("Failed to add wildcard record: %v", err)
+	}
+	
+	// Add exact record
+	exactIP := net.ParseIP("10.0.0.2")
+	err = store.AddRecord("exact.autoco.internal", exactIP)
+	if err != nil {
+		t.Fatalf("Failed to add exact record: %v", err)
+	}
+	
+	// Test exact match takes precedence
+	ips := store.GetRecords("exact.autoco.internal.", RecordTypeA)
+	if len(ips) != 1 {
+		t.Errorf("Expected 1 IP for exact match, got %d", len(ips))
+	}
+	if !ips[0].Equal(exactIP) {
+		t.Errorf("Expected exact IP %v, got %v", exactIP, ips[0])
+	}
+	
+	// Test wildcard match
+	ips = store.GetRecords("host.autoco.internal.", RecordTypeA)
+	if len(ips) != 1 {
+		t.Errorf("Expected 1 IP for wildcard match, got %d", len(ips))
+	}
+	if !ips[0].Equal(wildcardIP) {
+		t.Errorf("Expected wildcard IP %v, got %v", wildcardIP, ips[0])
+	}
+	
+	// Test non-match (base domain)
+	ips = store.GetRecords("autoco.internal.", RecordTypeA)
+	if len(ips) != 0 {
+		t.Errorf("Expected 0 IPs for base domain, got %d", len(ips))
+	}
+}
+
+func TestDNSRecordStoreComplexWildcard(t *testing.T) {
+	store := NewDNSRecordStore()
+	
+	// Add complex wildcard pattern
+	ip1 := net.ParseIP("10.0.0.1")
+	err := store.AddRecord("*.host-0?.autoco.internal", ip1)
+	if err != nil {
+		t.Fatalf("Failed to add wildcard record: %v", err)
+	}
+	
+	// Test matching domain
+	ips := store.GetRecords("sub.host-01.autoco.internal.", RecordTypeA)
+	if len(ips) != 1 {
+		t.Errorf("Expected 1 IP for complex wildcard match, got %d", len(ips))
+	}
+	if len(ips) > 0 && !ips[0].Equal(ip1) {
+		t.Errorf("Expected IP %v, got %v", ip1, ips[0])
+	}
+	
+	// Test non-matching domain (missing prefix)
+	ips = store.GetRecords("host-01.autoco.internal.", RecordTypeA)
+	if len(ips) != 0 {
+		t.Errorf("Expected 0 IPs for domain without prefix, got %d", len(ips))
+	}
+	
+	// Test non-matching domain (wrong ? position)
+	ips = store.GetRecords("sub.host-012.autoco.internal.", RecordTypeA)
+	if len(ips) != 0 {
+		t.Errorf("Expected 0 IPs for domain with wrong ? match, got %d", len(ips))
+	}
+}
+
+func TestDNSRecordStoreRemoveWildcard(t *testing.T) {
+	store := NewDNSRecordStore()
+	
+	// Add wildcard record
+	ip := net.ParseIP("10.0.0.1")
+	err := store.AddRecord("*.autoco.internal", ip)
+	if err != nil {
+		t.Fatalf("Failed to add wildcard record: %v", err)
+	}
+	
+	// Verify it exists
+	ips := store.GetRecords("host.autoco.internal.", RecordTypeA)
+	if len(ips) != 1 {
+		t.Errorf("Expected 1 IP before removal, got %d", len(ips))
+	}
+	
+	// Remove wildcard record
+	store.RemoveRecord("*.autoco.internal", nil)
+	
+	// Verify it's gone
+	ips = store.GetRecords("host.autoco.internal.", RecordTypeA)
+	if len(ips) != 0 {
+		t.Errorf("Expected 0 IPs after removal, got %d", len(ips))
+	}
+}
+
+func TestDNSRecordStoreMultipleWildcards(t *testing.T) {
+	store := NewDNSRecordStore()
+	
+	// Add multiple wildcard patterns that don't overlap
+	ip1 := net.ParseIP("10.0.0.1")
+	ip2 := net.ParseIP("10.0.0.2")
+	ip3 := net.ParseIP("10.0.0.3")
+	
+	err := store.AddRecord("*.prod.autoco.internal", ip1)
+	if err != nil {
+		t.Fatalf("Failed to add first wildcard: %v", err)
+	}
+	
+	err = store.AddRecord("*.dev.autoco.internal", ip2)
+	if err != nil {
+		t.Fatalf("Failed to add second wildcard: %v", err)
+	}
+	
+	// Add a broader wildcard that matches both
+	err = store.AddRecord("*.autoco.internal", ip3)
+	if err != nil {
+		t.Fatalf("Failed to add third wildcard: %v", err)
+	}
+	
+	// Test domain matching only the prod pattern and the broad pattern
+	ips := store.GetRecords("host.prod.autoco.internal.", RecordTypeA)
+	if len(ips) != 2 {
+		t.Errorf("Expected 2 IPs (prod + broad), got %d", len(ips))
+	}
+	
+	// Test domain matching only the dev pattern and the broad pattern
+	ips = store.GetRecords("service.dev.autoco.internal.", RecordTypeA)
+	if len(ips) != 2 {
+		t.Errorf("Expected 2 IPs (dev + broad), got %d", len(ips))
+	}
+	
+	// Test domain matching only the broad pattern
+	ips = store.GetRecords("host.test.autoco.internal.", RecordTypeA)
+	if len(ips) != 1 {
+		t.Errorf("Expected 1 IP (broad only), got %d", len(ips))
+	}
+}
+
+func TestDNSRecordStoreIPv6Wildcard(t *testing.T) {
+	store := NewDNSRecordStore()
+	
+	// Add IPv6 wildcard record
+	ip := net.ParseIP("2001:db8::1")
+	err := store.AddRecord("*.autoco.internal", ip)
+	if err != nil {
+		t.Fatalf("Failed to add IPv6 wildcard record: %v", err)
+	}
+	
+	// Test wildcard match for IPv6
+	ips := store.GetRecords("host.autoco.internal.", RecordTypeAAAA)
+	if len(ips) != 1 {
+		t.Errorf("Expected 1 IPv6 for wildcard match, got %d", len(ips))
+	}
+	if len(ips) > 0 && !ips[0].Equal(ip) {
+		t.Errorf("Expected IPv6 %v, got %v", ip, ips[0])
+	}
+}
+
+func TestHasRecordWildcard(t *testing.T) {
+	store := NewDNSRecordStore()
+	
+	// Add wildcard record
+	ip := net.ParseIP("10.0.0.1")
+	err := store.AddRecord("*.autoco.internal", ip)
+	if err != nil {
+		t.Fatalf("Failed to add wildcard record: %v", err)
+	}
+	
+	// Test HasRecord with wildcard match
+	if !store.HasRecord("host.autoco.internal.", RecordTypeA) {
+		t.Error("Expected HasRecord to return true for wildcard match")
+	}
+	
+	// Test HasRecord with non-match
+	if store.HasRecord("autoco.internal.", RecordTypeA) {
+		t.Error("Expected HasRecord to return false for base domain")
+	}
+}

dns/override/dns_override_darwin.go

@@ -0,0 +1,68 @@
+//go:build darwin && !ios
+
+package olm
+
+import (
+	"fmt"
+	"net/netip"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/olm/dns"
+	platform "github.com/fosrl/olm/dns/platform"
+)
+
+var configurator platform.DNSConfigurator
+
+// SetupDNSOverride configures the system DNS to use the DNS proxy on macOS
+// Uses scutil for DNS configuration
+func SetupDNSOverride(interfaceName string, dnsProxy *dns.DNSProxy) error {
+	if dnsProxy == nil {
+		return fmt.Errorf("DNS proxy is nil")
+	}
+
+	var err error
+	configurator, err = platform.NewDarwinDNSConfigurator()
+	if err != nil {
+		return fmt.Errorf("failed to create Darwin DNS configurator: %w", err)
+	}
+
+	logger.Info("Using Darwin scutil DNS configurator")
+
+	// Get current DNS servers before changing
+	currentDNS, err := configurator.GetCurrentDNS()
+	if err != nil {
+		logger.Warn("Could not get current DNS: %v", err)
+	} else {
+		logger.Info("Current DNS servers: %v", currentDNS)
+	}
+
+	// Set new DNS servers to point to our proxy
+	newDNS := []netip.Addr{
+		dnsProxy.GetProxyIP(),
+	}
+
+	logger.Info("Setting DNS servers to: %v", newDNS)
+	originalDNS, err := configurator.SetDNS(newDNS)
+	if err != nil {
+		return fmt.Errorf("failed to set DNS: %w", err)
+	}
+
+	logger.Info("Original DNS servers backed up: %v", originalDNS)
+	return nil
+}
+
+// RestoreDNSOverride restores the original DNS configuration
+func RestoreDNSOverride() error {
+	if configurator == nil {
+		logger.Debug("No DNS configurator to restore")
+		return nil
+	}
+
+	logger.Info("Restoring original DNS configuration")
+	if err := configurator.RestoreDNS(); err != nil {
+		return fmt.Errorf("failed to restore DNS: %w", err)
+	}
+
+	logger.Info("DNS configuration restored successfully")
+	return nil
+}

dns/override/dns_override_unix.go

@@ -0,0 +1,105 @@
+//go:build (linux && !android) || freebsd
+
+package olm
+
+import (
+	"fmt"
+	"net/netip"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/olm/dns"
+	platform "github.com/fosrl/olm/dns/platform"
+)
+
+var configurator platform.DNSConfigurator
+
+// SetupDNSOverride configures the system DNS to use the DNS proxy on Linux/FreeBSD
+// Detects the DNS manager by reading /etc/resolv.conf and verifying runtime availability
+func SetupDNSOverride(interfaceName string, dnsProxy *dns.DNSProxy) error {
+	if dnsProxy == nil {
+		return fmt.Errorf("DNS proxy is nil")
+	}
+
+	var err error
+
+	// Detect which DNS manager is in use by checking /etc/resolv.conf and runtime availability
+	managerType := platform.DetectDNSManager(interfaceName)
+	logger.Info("Detected DNS manager: %s", managerType.String())
+
+	// Create configurator based on detected manager
+	switch managerType {
+	case platform.SystemdResolvedManager:
+		configurator, err = platform.NewSystemdResolvedDNSConfigurator(interfaceName)
+		if err == nil {
+			logger.Info("Using systemd-resolved DNS configurator")
+			return setDNS(dnsProxy, configurator)
+		}
+		logger.Warn("Failed to create systemd-resolved configurator: %v, falling back", err)
+
+	case platform.NetworkManagerManager:
+		configurator, err = platform.NewNetworkManagerDNSConfigurator(interfaceName)
+		if err == nil {
+			logger.Info("Using NetworkManager DNS configurator")
+			return setDNS(dnsProxy, configurator)
+		}
+		logger.Warn("Failed to create NetworkManager configurator: %v, falling back", err)
+
+	case platform.ResolvconfManager:
+		configurator, err = platform.NewResolvconfDNSConfigurator(interfaceName)
+		if err == nil {
+			logger.Info("Using resolvconf DNS configurator")
+			return setDNS(dnsProxy, configurator)
+		}
+		logger.Warn("Failed to create resolvconf configurator: %v, falling back", err)
+	}
+
+	// Fall back to direct file manipulation
+	configurator, err = platform.NewFileDNSConfigurator()
+	if err != nil {
+		return fmt.Errorf("failed to create file DNS configurator: %w", err)
+	}
+
+	logger.Info("Using file-based DNS configurator")
+	return setDNS(dnsProxy, configurator)
+}
+
+// setDNS is a helper function to set DNS and log the results
+func setDNS(dnsProxy *dns.DNSProxy, conf platform.DNSConfigurator) error {
+	// Get current DNS servers before changing
+	currentDNS, err := conf.GetCurrentDNS()
+	if err != nil {
+		logger.Warn("Could not get current DNS: %v", err)
+	} else {
+		logger.Info("Current DNS servers: %v", currentDNS)
+	}
+
+	// Set new DNS servers to point to our proxy
+	newDNS := []netip.Addr{
+		dnsProxy.GetProxyIP(),
+	}
+
+	logger.Info("Setting DNS servers to: %v", newDNS)
+	originalDNS, err := conf.SetDNS(newDNS)
+	if err != nil {
+		return fmt.Errorf("failed to set DNS: %w", err)
+	}
+
+	logger.Info("Original DNS servers backed up: %v", originalDNS)
+	return nil
+}
+
+// RestoreDNSOverride restores the original DNS configuration
+func RestoreDNSOverride() error {
+	if configurator == nil {
+		logger.Debug("No DNS configurator to restore")
+		return nil
+	}
+
+	logger.Info("Restoring original DNS configuration")
+	if err := configurator.RestoreDNS(); err != nil {
+		return fmt.Errorf("failed to restore DNS: %w", err)
+	}
+
+	logger.Info("DNS configuration restored successfully")
+	return nil
+}

dns/override/dns_override_windows.go

@@ -0,0 +1,68 @@
+//go:build windows
+
+package olm
+
+import (
+	"fmt"
+	"net/netip"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/olm/dns"
+	platform "github.com/fosrl/olm/dns/platform"
+)
+
+var configurator platform.DNSConfigurator
+
+// SetupDNSOverride configures the system DNS to use the DNS proxy on Windows
+// Uses registry-based configuration (automatically extracts interface GUID)
+func SetupDNSOverride(interfaceName string, dnsProxy *dns.DNSProxy) error {
+	if dnsProxy == nil {
+		return fmt.Errorf("DNS proxy is nil")
+	}
+
+	var err error
+	configurator, err = platform.NewWindowsDNSConfigurator(interfaceName)
+	if err != nil {
+		return fmt.Errorf("failed to create Windows DNS configurator: %w", err)
+	}
+
+	logger.Info("Using Windows registry DNS configurator for interface: %s", interfaceName)
+
+	// Get current DNS servers before changing
+	currentDNS, err := configurator.GetCurrentDNS()
+	if err != nil {
+		logger.Warn("Could not get current DNS: %v", err)
+	} else {
+		logger.Info("Current DNS servers: %v", currentDNS)
+	}
+
+	// Set new DNS servers to point to our proxy
+	newDNS := []netip.Addr{
+		dnsProxy.GetProxyIP(),
+	}
+
+	logger.Info("Setting DNS servers to: %v", newDNS)
+	originalDNS, err := configurator.SetDNS(newDNS)
+	if err != nil {
+		return fmt.Errorf("failed to set DNS: %w", err)
+	}
+
+	logger.Info("Original DNS servers backed up: %v", originalDNS)
+	return nil
+}
+
+// RestoreDNSOverride restores the original DNS configuration
+func RestoreDNSOverride() error {
+	if configurator == nil {
+		logger.Debug("No DNS configurator to restore")
+		return nil
+	}
+
+	logger.Info("Restoring original DNS configuration")
+	if err := configurator.RestoreDNS(); err != nil {
+		return fmt.Errorf("failed to restore DNS: %w", err)
+	}
+
+	logger.Info("DNS configuration restored successfully")
+	return nil
+}

dns/platform/darwin.go

@@ -0,0 +1,419 @@
+//go:build darwin && !ios
+
+package dns
+
+import (
+	"bufio"
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"net/netip"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"runtime"
+	"strconv"
+	"strings"
+
+	"github.com/fosrl/newt/logger"
+)
+
+const (
+	scutilPath      = "/usr/sbin/scutil"
+	dscacheutilPath = "/usr/bin/dscacheutil"
+
+	dnsStateKeyFormat    = "State:/Network/Service/Olm-%s/DNS"
+	globalIPv4State      = "State:/Network/Global/IPv4"
+	primaryServiceFormat = "State:/Network/Service/%s/DNS"
+
+	keySupplementalMatchDomains         = "SupplementalMatchDomains"
+	keySupplementalMatchDomainsNoSearch = "SupplementalMatchDomainsNoSearch"
+	keyServerAddresses                  = "ServerAddresses"
+	keyServerPort                       = "ServerPort"
+	arraySymbol                         = "* "
+	digitSymbol                         = "# "
+
+	// State file name for crash recovery
+	dnsStateFileName = "dns_state.json"
+)
+
+// DNSPersistentState represents the state saved to disk for crash recovery
+type DNSPersistentState struct {
+	CreatedKeys []string `json:"created_keys"`
+}
+
+// DarwinDNSConfigurator manages DNS settings on macOS using scutil
+type DarwinDNSConfigurator struct {
+	createdKeys   map[string]struct{}
+	originalState *DNSState
+	stateFilePath string
+}
+
+// NewDarwinDNSConfigurator creates a new macOS DNS configurator
+func NewDarwinDNSConfigurator() (*DarwinDNSConfigurator, error) {
+	stateFilePath := getDNSStateFilePath()
+
+	configurator := &DarwinDNSConfigurator{
+		createdKeys:   make(map[string]struct{}),
+		stateFilePath: stateFilePath,
+	}
+
+	// Clean up any leftover state from a previous crash
+	if err := configurator.CleanupUncleanShutdown(); err != nil {
+		logger.Warn("Failed to cleanup previous DNS state: %v", err)
+	}
+
+	return configurator, nil
+}
+
+// Name returns the configurator name
+func (d *DarwinDNSConfigurator) Name() string {
+	return "darwin-scutil"
+}
+
+// SetDNS sets the DNS servers and returns the original servers
+func (d *DarwinDNSConfigurator) SetDNS(servers []netip.Addr) ([]netip.Addr, error) {
+	// Get current DNS settings before overriding
+	originalServers, err := d.GetCurrentDNS()
+	if err != nil {
+		return nil, fmt.Errorf("get current DNS: %w", err)
+	}
+
+	// Store original state
+	d.originalState = &DNSState{
+		OriginalServers:  originalServers,
+		ConfiguratorName: d.Name(),
+	}
+
+	// Set new DNS servers
+	if err := d.applyDNSServers(servers); err != nil {
+		return nil, fmt.Errorf("apply DNS servers: %w", err)
+	}
+
+	// Persist state to disk for crash recovery
+	if err := d.saveState(); err != nil {
+		logger.Warn("Failed to save DNS state for crash recovery: %v", err)
+	}
+
+	// Flush DNS cache
+	if err := d.flushDNSCache(); err != nil {
+		// Non-fatal, just log
+		fmt.Printf("warning: failed to flush DNS cache: %v\n", err)
+	}
+
+	return originalServers, nil
+}
+
+// RestoreDNS restores the original DNS configuration
+func (d *DarwinDNSConfigurator) RestoreDNS() error {
+	// Remove all created keys
+	for key := range d.createdKeys {
+		if err := d.removeKey(key); err != nil {
+			return fmt.Errorf("remove key %s: %w", key, err)
+		}
+	}
+
+	// Clear state file after successful restoration
+	if err := d.clearState(); err != nil {
+		logger.Warn("Failed to clear DNS state file: %v", err)
+	}
+
+	// Flush DNS cache
+	if err := d.flushDNSCache(); err != nil {
+		fmt.Printf("warning: failed to flush DNS cache: %v\n", err)
+	}
+
+	return nil
+}
+
+// GetCurrentDNS returns the currently configured DNS servers
+func (d *DarwinDNSConfigurator) GetCurrentDNS() ([]netip.Addr, error) {
+	primaryServiceKey, err := d.getPrimaryServiceKey()
+	if err != nil || primaryServiceKey == "" {
+		return nil, fmt.Errorf("get primary service: %w", err)
+	}
+
+	dnsKey := fmt.Sprintf(primaryServiceFormat, primaryServiceKey)
+	cmd := fmt.Sprintf("show %s\n", dnsKey)
+
+	output, err := d.runScutil(cmd)
+	if err != nil {
+		return nil, fmt.Errorf("run scutil: %w", err)
+	}
+
+	servers := d.parseServerAddresses(output)
+	return servers, nil
+}
+
+// CleanupUncleanShutdown removes any DNS keys left over from a previous crash
+func (d *DarwinDNSConfigurator) CleanupUncleanShutdown() error {
+	state, err := d.loadState()
+	if err != nil {
+		if os.IsNotExist(err) {
+			// No state file, nothing to clean up
+			return nil
+		}
+		return fmt.Errorf("load state: %w", err)
+	}
+
+	if len(state.CreatedKeys) == 0 {
+		// No keys to clean up
+		return nil
+	}
+
+	logger.Info("Found DNS state from previous session, cleaning up %d keys", len(state.CreatedKeys))
+
+	// Remove all keys from previous session
+	var lastErr error
+	for _, key := range state.CreatedKeys {
+		logger.Debug("Removing leftover DNS key: %s", key)
+		if err := d.removeKeyDirect(key); err != nil {
+			logger.Warn("Failed to remove DNS key %s: %v", key, err)
+			lastErr = err
+		}
+	}
+
+	// Clear state file
+	if err := d.clearState(); err != nil {
+		logger.Warn("Failed to clear DNS state file: %v", err)
+	}
+
+	// Flush DNS cache after cleanup
+	if err := d.flushDNSCache(); err != nil {
+		logger.Warn("Failed to flush DNS cache after cleanup: %v", err)
+	}
+
+	return lastErr
+}
+
+// applyDNSServers applies the DNS server configuration
+func (d *DarwinDNSConfigurator) applyDNSServers(servers []netip.Addr) error {
+	if len(servers) == 0 {
+		return fmt.Errorf("no DNS servers provided")
+	}
+
+	key := fmt.Sprintf(dnsStateKeyFormat, "Override")
+
+	// Use SupplementalMatchDomains with empty string to match ALL domains
+	// This is the key to making DNS override work on macOS
+	// Setting SupplementalMatchDomainsNoSearch to 0 enables search domain behavior
+	err := d.addDNSState(key, "\"\"", servers[0], 53, true)
+	if err != nil {
+		return fmt.Errorf("set DNS servers: %w", err)
+	}
+
+	d.createdKeys[key] = struct{}{}
+	return nil
+}
+
+// addDNSState adds a DNS state entry with the specified configuration
+func (d *DarwinDNSConfigurator) addDNSState(state, domains string, dnsServer netip.Addr, port int, enableSearch bool) error {
+	noSearch := "1"
+	if enableSearch {
+		noSearch = "0"
+	}
+
+	// Build the scutil command following NetBird's approach
+	var commands strings.Builder
+	commands.WriteString("d.init\n")
+	commands.WriteString(fmt.Sprintf("d.add %s %s%s\n", keySupplementalMatchDomains, arraySymbol, domains))
+	commands.WriteString(fmt.Sprintf("d.add %s %s%s\n", keySupplementalMatchDomainsNoSearch, digitSymbol, noSearch))
+	commands.WriteString(fmt.Sprintf("d.add %s %s%s\n", keyServerAddresses, arraySymbol, dnsServer.String()))
+	commands.WriteString(fmt.Sprintf("d.add %s %s%s\n", keyServerPort, digitSymbol, strconv.Itoa(port)))
+	commands.WriteString(fmt.Sprintf("set %s\n", state))
+
+	if _, err := d.runScutil(commands.String()); err != nil {
+		return fmt.Errorf("applying state for domains %s, error: %w", domains, err)
+	}
+
+	logger.Info("Added DNS override with server %s:%d for domains: %s", dnsServer.String(), port, domains)
+	return nil
+}
+
+// removeKey removes a DNS configuration key and updates internal state
+func (d *DarwinDNSConfigurator) removeKey(key string) error {
+	if err := d.removeKeyDirect(key); err != nil {
+		return err
+	}
+
+	delete(d.createdKeys, key)
+	return nil
+}
+
+// removeKeyDirect removes a DNS configuration key without updating internal state
+// Used for cleanup operations
+func (d *DarwinDNSConfigurator) removeKeyDirect(key string) error {
+	cmd := fmt.Sprintf("remove %s\n", key)
+
+	if _, err := d.runScutil(cmd); err != nil {
+		return fmt.Errorf("remove key: %w", err)
+	}
+
+	return nil
+}
+
+// getPrimaryServiceKey gets the primary network service key
+func (d *DarwinDNSConfigurator) getPrimaryServiceKey() (string, error) {
+	cmd := fmt.Sprintf("show %s\n", globalIPv4State)
+
+	output, err := d.runScutil(cmd)
+	if err != nil {
+		return "", fmt.Errorf("run scutil: %w", err)
+	}
+
+	scanner := bufio.NewScanner(bytes.NewReader(output))
+	for scanner.Scan() {
+		line := scanner.Text()
+		if strings.Contains(line, "PrimaryService") {
+			parts := strings.Split(line, ":")
+			if len(parts) >= 2 {
+				return strings.TrimSpace(parts[1]), nil
+			}
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		return "", fmt.Errorf("scan output: %w", err)
+	}
+
+	return "", fmt.Errorf("primary service not found")
+}
+
+// parseServerAddresses parses DNS server addresses from scutil output
+func (d *DarwinDNSConfigurator) parseServerAddresses(output []byte) []netip.Addr {
+	var servers []netip.Addr
+	inServerArray := false
+
+	scanner := bufio.NewScanner(bytes.NewReader(output))
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+
+		if strings.HasPrefix(line, "ServerAddresses : <array> {") {
+			inServerArray = true
+			continue
+		}
+
+		if line == "}" {
+			inServerArray = false
+			continue
+		}
+
+		if inServerArray {
+			// Line format: "0 : 8.8.8.8"
+			parts := strings.Split(line, " : ")
+			if len(parts) >= 2 {
+				if addr, err := netip.ParseAddr(parts[1]); err == nil {
+					servers = append(servers, addr)
+				}
+			}
+		}
+	}
+
+	return servers
+}
+
+// flushDNSCache flushes the system DNS cache
+func (d *DarwinDNSConfigurator) flushDNSCache() error {
+	logger.Debug("Flushing dscacheutil cache")
+	cmd := exec.Command(dscacheutilPath, "-flushcache")
+	if err := cmd.Run(); err != nil {
+		return fmt.Errorf("flush cache: %w", err)
+	}
+
+	logger.Debug("Flushing mDNSResponder cache")
+
+	cmd = exec.Command("killall", "-HUP", "mDNSResponder")
+	if err := cmd.Run(); err != nil {
+		// Non-fatal, mDNSResponder might not be running
+		return nil
+	}
+
+	return nil
+}
+
+// runScutil executes an scutil command
+func (d *DarwinDNSConfigurator) runScutil(commands string) ([]byte, error) {
+	// Wrap commands with open/quit
+	wrapped := fmt.Sprintf("open\n%squit\n", commands)
+
+	logger.Debug("Running scutil with commands:\n%s\n", wrapped)
+
+	cmd := exec.Command(scutilPath)
+	cmd.Stdin = strings.NewReader(wrapped)
+
+	output, err := cmd.CombinedOutput()
+	if err != nil {
+		return nil, fmt.Errorf("scutil command failed: %w, output: %s", err, output)
+	}
+
+	logger.Debug("scutil output:\n%s\n", output)
+
+	return output, nil
+}
+
+// getDNSStateFilePath returns the path to the DNS state file
+func getDNSStateFilePath() string {
+	var stateDir string
+	switch runtime.GOOS {
+	case "darwin":
+		stateDir = filepath.Join(os.Getenv("HOME"), "Library", "Application Support", "olm-client")
+	default:
+		stateDir = filepath.Join(os.Getenv("HOME"), ".config", "olm-client")
+	}
+
+	if err := os.MkdirAll(stateDir, 0755); err != nil {
+		logger.Warn("Failed to create state directory: %v", err)
+	}
+
+	return filepath.Join(stateDir, dnsStateFileName)
+}
+
+// saveState persists the current DNS state to disk
+func (d *DarwinDNSConfigurator) saveState() error {
+	keys := make([]string, 0, len(d.createdKeys))
+	for key := range d.createdKeys {
+		keys = append(keys, key)
+	}
+
+	state := DNSPersistentState{
+		CreatedKeys: keys,
+	}
+
+	data, err := json.MarshalIndent(state, "", "  ")
+	if err != nil {
+		return fmt.Errorf("marshal state: %w", err)
+	}
+
+	if err := os.WriteFile(d.stateFilePath, data, 0644); err != nil {
+		return fmt.Errorf("write state file: %w", err)
+	}
+
+	logger.Debug("Saved DNS state to %s", d.stateFilePath)
+	return nil
+}
+
+// loadState loads the DNS state from disk
+func (d *DarwinDNSConfigurator) loadState() (*DNSPersistentState, error) {
+	data, err := os.ReadFile(d.stateFilePath)
+	if err != nil {
+		return nil, err
+	}
+
+	var state DNSPersistentState
+	if err := json.Unmarshal(data, &state); err != nil {
+		return nil, fmt.Errorf("unmarshal state: %w", err)
+	}
+
+	return &state, nil
+}
+
+// clearState removes the DNS state file
+func (d *DarwinDNSConfigurator) clearState() error {
+	err := os.Remove(d.stateFilePath)
+	if err != nil && !os.IsNotExist(err) {
+		return fmt.Errorf("remove state file: %w", err)
+	}
+
+	logger.Debug("Cleared DNS state file")
+	return nil
+}

dns/platform/detect_unix.go

@@ -0,0 +1,158 @@
+//go:build (linux && !android) || freebsd
+
+package dns
+
+import (
+	"bufio"
+	"io"
+	"os"
+	"strings"
+
+	"github.com/fosrl/newt/logger"
+)
+
+const defaultResolvConfPath = "/etc/resolv.conf"
+
+// DNSManagerType represents the type of DNS manager detected
+type DNSManagerType int
+
+const (
+	// UnknownManager indicates we couldn't determine the DNS manager
+	UnknownManager DNSManagerType = iota
+	// SystemdResolvedManager indicates systemd-resolved is managing DNS
+	SystemdResolvedManager
+	// NetworkManagerManager indicates NetworkManager is managing DNS
+	NetworkManagerManager
+	// ResolvconfManager indicates resolvconf is managing DNS
+	ResolvconfManager
+	// FileManager indicates direct file management (no DNS manager)
+	FileManager
+)
+
+// DetectDNSManagerFromFile reads /etc/resolv.conf to determine which DNS manager is in use
+// This provides a hint based on comments in the file, similar to Netbird's approach
+func DetectDNSManagerFromFile() DNSManagerType {
+	file, err := os.Open(defaultResolvConfPath)
+	if err != nil {
+		return UnknownManager
+	}
+	defer file.Close()
+
+	scanner := bufio.NewScanner(file)
+	for scanner.Scan() {
+		text := scanner.Text()
+		if len(text) == 0 {
+			continue
+		}
+
+		// If we hit a non-comment line, default to file-based
+		if text[0] != '#' {
+			return FileManager
+		}
+
+		// Check for DNS manager signatures in comments
+		if strings.Contains(text, "NetworkManager") {
+			return NetworkManagerManager
+		}
+
+		if strings.Contains(text, "systemd-resolved") {
+			return SystemdResolvedManager
+		}
+
+		if strings.Contains(text, "resolvconf") {
+			return ResolvconfManager
+		}
+	}
+
+	if err := scanner.Err(); err != nil && err != io.EOF {
+		return UnknownManager
+	}
+
+	// No indicators found, assume file-based management
+	return FileManager
+}
+
+// String returns a human-readable name for the DNS manager type
+func (d DNSManagerType) String() string {
+	switch d {
+	case SystemdResolvedManager:
+		return "systemd-resolved"
+	case NetworkManagerManager:
+		return "NetworkManager"
+	case ResolvconfManager:
+		return "resolvconf"
+	case FileManager:
+		return "file"
+	default:
+		return "unknown"
+	}
+}
+
+// DetectDNSManager combines file detection with runtime availability checks
+// to determine the best DNS configurator to use
+func DetectDNSManager(interfaceName string) DNSManagerType {
+	// First check what the file suggests
+	fileHint := DetectDNSManagerFromFile()
+
+	// Verify the hint with runtime checks
+	switch fileHint {
+	case SystemdResolvedManager:
+		// Verify systemd-resolved is actually running
+		if IsSystemdResolvedAvailable() {
+			return SystemdResolvedManager
+		}
+		logger.Warn("dns platform: Found systemd-resolved but it is not running. Falling back to file...")
+		os.Exit(0)
+		return FileManager
+
+	case NetworkManagerManager:
+		// Verify NetworkManager is actually running
+		if IsNetworkManagerAvailable() {
+			// Check if NetworkManager is delegating to systemd-resolved
+			if !IsNetworkManagerDNSModeSupported() {
+				logger.Info("NetworkManager is delegating DNS to systemd-resolved, using systemd-resolved configurator")
+				if IsSystemdResolvedAvailable() {
+					return SystemdResolvedManager
+				}
+			}
+			return NetworkManagerManager
+		}
+		logger.Warn("dns platform: Found network manager but it is not running. Falling back to file...")
+		return FileManager
+
+	case ResolvconfManager:
+		// Verify resolvconf is available
+		if IsResolvconfAvailable() {
+			return ResolvconfManager
+		}
+		// If resolvconf is mentioned but not available, fall back to file
+		return FileManager
+
+	case FileManager:
+		// File suggests direct file management
+		// But we should still check if a manager is available that wasn't mentioned
+		if IsSystemdResolvedAvailable() && interfaceName != "" {
+			return SystemdResolvedManager
+		}
+		if IsNetworkManagerAvailable() && interfaceName != "" {
+			return NetworkManagerManager
+		}
+		if IsResolvconfAvailable() && interfaceName != "" {
+			return ResolvconfManager
+		}
+		return FileManager
+
+	default:
+		// Unknown - do runtime detection
+		if IsSystemdResolvedAvailable() && interfaceName != "" {
+			return SystemdResolvedManager
+		}
+		if IsNetworkManagerAvailable() && interfaceName != "" {
+			return NetworkManagerManager
+		}
+		if IsResolvconfAvailable() && interfaceName != "" {
+			return ResolvconfManager
+		}
+		return FileManager
+	}
+}

dns/platform/file.go

@@ -0,0 +1,220 @@
+//go:build (linux && !android) || freebsd
+
+package dns
+
+import (
+	"fmt"
+	"net/netip"
+	"os"
+	"strings"
+)
+
+const (
+	resolvConfPath       = "/etc/resolv.conf"
+	resolvConfBackupPath = "/etc/resolv.conf.olm.backup"
+	resolvConfHeader     = "# Generated by Olm DNS Manager\n# Original file backed up to " + resolvConfBackupPath + "\n\n"
+)
+
+// FileDNSConfigurator manages DNS settings by directly modifying /etc/resolv.conf
+type FileDNSConfigurator struct {
+	originalState *DNSState
+}
+
+// NewFileDNSConfigurator creates a new file-based DNS configurator
+func NewFileDNSConfigurator() (*FileDNSConfigurator, error) {
+	f := &FileDNSConfigurator{}
+	if err := f.CleanupUncleanShutdown(); err != nil {
+		return nil, fmt.Errorf("cleanup unclean shutdown: %w", err)
+	}
+	return f, nil
+}
+
+// Name returns the configurator name
+func (f *FileDNSConfigurator) Name() string {
+	return "file-resolv.conf"
+}
+
+// SetDNS sets the DNS servers and returns the original servers
+func (f *FileDNSConfigurator) SetDNS(servers []netip.Addr) ([]netip.Addr, error) {
+	// Get current DNS settings before overriding
+	originalServers, err := f.GetCurrentDNS()
+	if err != nil {
+		return nil, fmt.Errorf("get current DNS: %w", err)
+	}
+
+	// Backup original resolv.conf if not already backed up
+	if !f.isBackupExists() {
+		if err := f.backupResolvConf(); err != nil {
+			return nil, fmt.Errorf("backup resolv.conf: %w", err)
+		}
+	}
+
+	// Store original state
+	f.originalState = &DNSState{
+		OriginalServers:  originalServers,
+		ConfiguratorName: f.Name(),
+	}
+
+	// Write new resolv.conf
+	if err := f.writeResolvConf(servers); err != nil {
+		return nil, fmt.Errorf("write resolv.conf: %w", err)
+	}
+
+	return originalServers, nil
+}
+
+// RestoreDNS restores the original DNS configuration
+func (f *FileDNSConfigurator) RestoreDNS() error {
+	if !f.isBackupExists() {
+		return fmt.Errorf("no backup file exists")
+	}
+
+	// Copy backup back to original location
+	if err := copyFile(resolvConfBackupPath, resolvConfPath); err != nil {
+		return fmt.Errorf("restore from backup: %w", err)
+	}
+
+	// Remove backup file
+	if err := os.Remove(resolvConfBackupPath); err != nil {
+		return fmt.Errorf("remove backup file: %w", err)
+	}
+
+	return nil
+}
+
+// CleanupUncleanShutdown removes any DNS configuration left over from a previous crash
+// For the file-based configurator, we check if a backup file exists (indicating a crash
+// happened while DNS was configured) and restore from it if so.
+func (f *FileDNSConfigurator) CleanupUncleanShutdown() error {
+	// Check if backup file exists from a previous session
+	if !f.isBackupExists() {
+		// No backup file, nothing to clean up
+		return nil
+	}
+
+	// A backup exists, which means we crashed while DNS was configured
+	// Restore the original resolv.conf
+	if err := copyFile(resolvConfBackupPath, resolvConfPath); err != nil {
+		return fmt.Errorf("restore from backup during cleanup: %w", err)
+	}
+
+	// Remove backup file
+	if err := os.Remove(resolvConfBackupPath); err != nil {
+		return fmt.Errorf("remove backup file during cleanup: %w", err)
+	}
+
+	return nil
+}
+
+// GetCurrentDNS returns the currently configured DNS servers
+func (f *FileDNSConfigurator) GetCurrentDNS() ([]netip.Addr, error) {
+	content, err := os.ReadFile(resolvConfPath)
+	if err != nil {
+		return nil, fmt.Errorf("read resolv.conf: %w", err)
+	}
+
+	return f.parseNameservers(string(content)), nil
+}
+
+// backupResolvConf creates a backup of the current resolv.conf
+func (f *FileDNSConfigurator) backupResolvConf() error {
+	// Get file info for permissions
+	info, err := os.Stat(resolvConfPath)
+	if err != nil {
+		return fmt.Errorf("stat resolv.conf: %w", err)
+	}
+
+	if err := copyFile(resolvConfPath, resolvConfBackupPath); err != nil {
+		return fmt.Errorf("copy file: %w", err)
+	}
+
+	// Preserve permissions
+	if err := os.Chmod(resolvConfBackupPath, info.Mode()); err != nil {
+		return fmt.Errorf("chmod backup: %w", err)
+	}
+
+	return nil
+}
+
+// writeResolvConf writes a new resolv.conf with the specified DNS servers
+func (f *FileDNSConfigurator) writeResolvConf(servers []netip.Addr) error {
+	if len(servers) == 0 {
+		return fmt.Errorf("no DNS servers provided")
+	}
+
+	// Get file info for permissions
+	info, err := os.Stat(resolvConfPath)
+	if err != nil {
+		return fmt.Errorf("stat resolv.conf: %w", err)
+	}
+
+	var content strings.Builder
+	content.WriteString(resolvConfHeader)
+
+	// Write nameservers
+	for _, server := range servers {
+		content.WriteString("nameserver ")
+		content.WriteString(server.String())
+		content.WriteString("\n")
+	}
+
+	// Write the file
+	if err := os.WriteFile(resolvConfPath, []byte(content.String()), info.Mode()); err != nil {
+		return fmt.Errorf("write resolv.conf: %w", err)
+	}
+
+	return nil
+}
+
+// isBackupExists checks if a backup file exists
+func (f *FileDNSConfigurator) isBackupExists() bool {
+	_, err := os.Stat(resolvConfBackupPath)
+	return err == nil
+}
+
+// parseNameservers extracts nameserver entries from resolv.conf content
+func (f *FileDNSConfigurator) parseNameservers(content string) []netip.Addr {
+	var servers []netip.Addr
+
+	lines := strings.Split(content, "\n")
+	for _, line := range lines {
+		line = strings.TrimSpace(line)
+
+		// Skip comments and empty lines
+		if line == "" || strings.HasPrefix(line, "#") {
+			continue
+		}
+
+		// Look for nameserver lines
+		if strings.HasPrefix(line, "nameserver") {
+			fields := strings.Fields(line)
+			if len(fields) >= 2 {
+				if addr, err := netip.ParseAddr(fields[1]); err == nil {
+					servers = append(servers, addr)
+				}
+			}
+		}
+	}
+
+	return servers
+}
+
+// copyFile copies a file from src to dst
+func copyFile(src, dst string) error {
+	content, err := os.ReadFile(src)
+	if err != nil {
+		return fmt.Errorf("read source: %w", err)
+	}
+
+	// Get source file permissions
+	info, err := os.Stat(src)
+	if err != nil {
+		return fmt.Errorf("stat source: %w", err)
+	}
+
+	if err := os.WriteFile(dst, content, info.Mode()); err != nil {
+		return fmt.Errorf("write destination: %w", err)
+	}
+
+	return nil
+}

dns/platform/network_manager.go

@@ -0,0 +1,325 @@
+//go:build (linux && !android) || freebsd
+
+package dns
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"net/netip"
+	"os"
+	"strings"
+	"time"
+
+	dbus "github.com/godbus/dbus/v5"
+)
+
+const (
+	// NetworkManager D-Bus constants
+	networkManagerDest                       = "org.freedesktop.NetworkManager"
+	networkManagerDbusObjectNode             = "/org/freedesktop/NetworkManager"
+	networkManagerDbusDNSManagerInterface    = "org.freedesktop.NetworkManager.DnsManager"
+	networkManagerDbusDNSManagerObjectNode   = networkManagerDbusObjectNode + "/DnsManager"
+	networkManagerDbusDNSManagerModeProperty = networkManagerDbusDNSManagerInterface + ".Mode"
+	networkManagerDbusVersionProperty        = "org.freedesktop.NetworkManager.Version"
+
+	// NetworkManager dispatcher script path
+	networkManagerDispatcherDir  = "/etc/NetworkManager/dispatcher.d"
+	networkManagerConfDir        = "/etc/NetworkManager/conf.d"
+	networkManagerDNSConfFile    = "olm-dns.conf"
+	networkManagerDispatcherFile = "01-olm-dns"
+)
+
+// NetworkManagerDNSConfigurator manages DNS settings using NetworkManager configuration files
+// This approach works with unmanaged interfaces by modifying NetworkManager's global DNS settings
+type NetworkManagerDNSConfigurator struct {
+	ifaceName     string
+	originalState *DNSState
+	confPath      string
+	dispatchPath  string
+}
+
+// NewNetworkManagerDNSConfigurator creates a new NetworkManager DNS configurator
+func NewNetworkManagerDNSConfigurator(ifaceName string) (*NetworkManagerDNSConfigurator, error) {
+	if ifaceName == "" {
+		return nil, fmt.Errorf("interface name is required")
+	}
+
+	// Check that NetworkManager conf.d directory exists
+	if _, err := os.Stat(networkManagerConfDir); os.IsNotExist(err) {
+		return nil, fmt.Errorf("NetworkManager conf.d directory not found: %s", networkManagerConfDir)
+	}
+
+	configurator := &NetworkManagerDNSConfigurator{
+		ifaceName:    ifaceName,
+		confPath:     networkManagerConfDir + "/" + networkManagerDNSConfFile,
+		dispatchPath: networkManagerDispatcherDir + "/" + networkManagerDispatcherFile,
+	}
+
+	// Clean up any stale configuration from a previous unclean shutdown
+	if err := configurator.CleanupUncleanShutdown(); err != nil {
+		return nil, fmt.Errorf("cleanup unclean shutdown: %w", err)
+	}
+
+	return configurator, nil
+}
+
+// Name returns the configurator name
+func (n *NetworkManagerDNSConfigurator) Name() string {
+	return "network-manager"
+}
+
+// SetDNS sets the DNS servers and returns the original servers
+func (n *NetworkManagerDNSConfigurator) SetDNS(servers []netip.Addr) ([]netip.Addr, error) {
+	// Get current DNS settings before overriding
+	originalServers, err := n.GetCurrentDNS()
+	if err != nil {
+		// If we can't get current DNS, proceed anyway
+		originalServers = []netip.Addr{}
+	}
+
+	// Store original state
+	n.originalState = &DNSState{
+		OriginalServers:  originalServers,
+		ConfiguratorName: n.Name(),
+	}
+
+	// Apply new DNS servers
+	if err := n.applyDNSServers(servers); err != nil {
+		return nil, fmt.Errorf("apply DNS servers: %w", err)
+	}
+
+	return originalServers, nil
+}
+
+// RestoreDNS restores the original DNS configuration
+func (n *NetworkManagerDNSConfigurator) RestoreDNS() error {
+	// Remove our configuration file
+	if err := os.Remove(n.confPath); err != nil && !os.IsNotExist(err) {
+		return fmt.Errorf("remove DNS config file: %w", err)
+	}
+
+	// Reload NetworkManager to apply the change
+	if err := n.reloadNetworkManager(); err != nil {
+		return fmt.Errorf("reload NetworkManager: %w", err)
+	}
+
+	return nil
+}
+
+// CleanupUncleanShutdown removes any DNS configuration left over from a previous crash
+// For NetworkManager, we check if our config file exists and remove it if so.
+// This ensures that if the process crashed while DNS was configured, the stale
+// configuration is removed on the next startup.
+func (n *NetworkManagerDNSConfigurator) CleanupUncleanShutdown() error {
+	// Check if our config file exists from a previous session
+	if _, err := os.Stat(n.confPath); os.IsNotExist(err) {
+		// No config file, nothing to clean up
+		return nil
+	}
+
+	// Remove the stale configuration file
+	if err := os.Remove(n.confPath); err != nil && !os.IsNotExist(err) {
+		return fmt.Errorf("remove stale DNS config file: %w", err)
+	}
+
+	// Reload NetworkManager to apply the change
+	if err := n.reloadNetworkManager(); err != nil {
+		return fmt.Errorf("reload NetworkManager after cleanup: %w", err)
+	}
+
+	return nil
+}
+
+// GetCurrentDNS returns the currently configured DNS servers by reading /etc/resolv.conf
+func (n *NetworkManagerDNSConfigurator) GetCurrentDNS() ([]netip.Addr, error) {
+	content, err := os.ReadFile("/etc/resolv.conf")
+	if err != nil {
+		return nil, fmt.Errorf("read resolv.conf: %w", err)
+	}
+
+	var servers []netip.Addr
+	lines := strings.Split(string(content), "\n")
+	for _, line := range lines {
+		line = strings.TrimSpace(line)
+		if strings.HasPrefix(line, "nameserver") {
+			fields := strings.Fields(line)
+			if len(fields) >= 2 {
+				if addr, err := netip.ParseAddr(fields[1]); err == nil {
+					servers = append(servers, addr)
+				}
+			}
+		}
+	}
+
+	return servers, nil
+}
+
+// applyDNSServers applies DNS server configuration via NetworkManager config file
+func (n *NetworkManagerDNSConfigurator) applyDNSServers(servers []netip.Addr) error {
+	if len(servers) == 0 {
+		return fmt.Errorf("no DNS servers provided")
+	}
+
+	// Build DNS server list
+	var dnsServers []string
+	for _, server := range servers {
+		dnsServers = append(dnsServers, server.String())
+	}
+
+	// Create NetworkManager configuration file that sets global DNS
+	// This overrides DNS for all connections
+	configContent := fmt.Sprintf(`# Generated by Olm DNS Manager - DO NOT EDIT
+# This file configures NetworkManager to use Olm's DNS proxy
+
+[global-dns-domain-*]
+servers=%s
+`, strings.Join(dnsServers, ","))
+
+	// Write the configuration file
+	if err := os.WriteFile(n.confPath, []byte(configContent), 0644); err != nil {
+		return fmt.Errorf("write DNS config file: %w", err)
+	}
+
+	// Reload NetworkManager to apply the new configuration
+	if err := n.reloadNetworkManager(); err != nil {
+		// Try to clean up
+		os.Remove(n.confPath)
+		return fmt.Errorf("reload NetworkManager: %w", err)
+	}
+
+	return nil
+}
+
+// reloadNetworkManager tells NetworkManager to reload its configuration
+func (n *NetworkManagerDNSConfigurator) reloadNetworkManager() error {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(networkManagerDest, networkManagerDbusObjectNode)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	defer cancel()
+
+	// Call Reload method with flags=0 (reload everything)
+	// See: https://networkmanager.dev/docs/api/latest/gdbus-org.freedesktop.NetworkManager.html#gdbus-method-org-freedesktop-NetworkManager.Reload
+	err = obj.CallWithContext(ctx, networkManagerDest+".Reload", 0, uint32(0)).Store()
+	if err != nil {
+		return fmt.Errorf("call Reload: %w", err)
+	}
+
+	return nil
+}
+
+// IsNetworkManagerAvailable checks if NetworkManager is available and responsive
+func IsNetworkManagerAvailable() bool {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return false
+	}
+	defer conn.Close()
+
+	obj := conn.Object(networkManagerDest, networkManagerDbusObjectNode)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+
+	// Try to ping NetworkManager
+	if err := obj.CallWithContext(ctx, "org.freedesktop.DBus.Peer.Ping", 0).Store(); err != nil {
+		return false
+	}
+
+	return true
+}
+
+// IsNetworkManagerDNSModeSupported checks if NetworkManager's DNS mode is one we can work with
+// Some DNS modes delegate to other systems (like systemd-resolved) which we should use directly
+func IsNetworkManagerDNSModeSupported() bool {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return false
+	}
+	defer conn.Close()
+
+	obj := conn.Object(networkManagerDest, networkManagerDbusDNSManagerObjectNode)
+
+	modeVariant, err := obj.GetProperty(networkManagerDbusDNSManagerModeProperty)
+	if err != nil {
+		// If we can't get the mode, assume it's not supported
+		return false
+	}
+
+	mode, ok := modeVariant.Value().(string)
+	if !ok {
+		return false
+	}
+
+	// If NetworkManager is delegating DNS to systemd-resolved, we should use
+	// systemd-resolved directly for better control
+	switch mode {
+	case "systemd-resolved":
+		// NetworkManager is delegating to systemd-resolved
+		// We should use systemd-resolved configurator instead
+		return false
+	case "dnsmasq", "unbound":
+		// NetworkManager is using a local resolver that it controls
+		// We can configure DNS through NetworkManager
+		return true
+	case "default", "none", "":
+		// NetworkManager is managing DNS directly or not at all
+		// We can configure DNS through NetworkManager
+		return true
+	default:
+		// Unknown mode, try to use it
+		return true
+	}
+}
+
+// GetNetworkManagerDNSMode returns the current DNS mode of NetworkManager
+func GetNetworkManagerDNSMode() (string, error) {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return "", fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(networkManagerDest, networkManagerDbusDNSManagerObjectNode)
+
+	modeVariant, err := obj.GetProperty(networkManagerDbusDNSManagerModeProperty)
+	if err != nil {
+		return "", fmt.Errorf("get DNS mode property: %w", err)
+	}
+
+	mode, ok := modeVariant.Value().(string)
+	if !ok {
+		return "", errors.New("DNS mode is not a string")
+	}
+
+	return mode, nil
+}
+
+// GetNetworkManagerVersion returns the version of NetworkManager
+func GetNetworkManagerVersion() (string, error) {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return "", fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(networkManagerDest, networkManagerDbusObjectNode)
+
+	versionVariant, err := obj.GetProperty(networkManagerDbusVersionProperty)
+	if err != nil {
+		return "", fmt.Errorf("get version property: %w", err)
+	}
+
+	version, ok := versionVariant.Value().(string)
+	if !ok {
+		return "", errors.New("version is not a string")
+	}
+
+	return version, nil
+}

dns/platform/resolvconf.go

@@ -0,0 +1,221 @@
+//go:build (linux && !android) || freebsd
+
+package dns
+
+import (
+	"bytes"
+	"fmt"
+	"net/netip"
+	"os/exec"
+	"strings"
+)
+
+const resolvconfCommand = "resolvconf"
+
+// ResolvconfDNSConfigurator manages DNS settings using the resolvconf utility
+type ResolvconfDNSConfigurator struct {
+	ifaceName     string
+	implType      string
+	originalState *DNSState
+}
+
+// NewResolvconfDNSConfigurator creates a new resolvconf DNS configurator
+func NewResolvconfDNSConfigurator(ifaceName string) (*ResolvconfDNSConfigurator, error) {
+	if ifaceName == "" {
+		return nil, fmt.Errorf("interface name is required")
+	}
+
+	// Detect resolvconf implementation type
+	implType, err := detectResolvconfType()
+	if err != nil {
+		return nil, fmt.Errorf("detect resolvconf type: %w", err)
+	}
+
+	configurator := &ResolvconfDNSConfigurator{
+		ifaceName: ifaceName,
+		implType:  implType,
+	}
+
+	// Call cleanup function to remove any stale DNS config for this interface
+	if err := configurator.CleanupUncleanShutdown(); err != nil {
+		return nil, fmt.Errorf("cleanup unclean shutdown: %w", err)
+	}
+
+	return configurator, nil
+}
+
+// Name returns the configurator name
+func (r *ResolvconfDNSConfigurator) Name() string {
+	return fmt.Sprintf("resolvconf-%s", r.implType)
+}
+
+// SetDNS sets the DNS servers and returns the original servers
+func (r *ResolvconfDNSConfigurator) SetDNS(servers []netip.Addr) ([]netip.Addr, error) {
+	// Get current DNS settings before overriding
+	originalServers, err := r.GetCurrentDNS()
+	if err != nil {
+		// If we can't get current DNS, proceed anyway
+		originalServers = []netip.Addr{}
+	}
+
+	// Store original state
+	r.originalState = &DNSState{
+		OriginalServers:  originalServers,
+		ConfiguratorName: r.Name(),
+	}
+
+	// Apply new DNS servers
+	if err := r.applyDNSServers(servers); err != nil {
+		return nil, fmt.Errorf("apply DNS servers: %w", err)
+	}
+
+	return originalServers, nil
+}
+
+// RestoreDNS restores the original DNS configuration
+func (r *ResolvconfDNSConfigurator) RestoreDNS() error {
+	var cmd *exec.Cmd
+
+	switch r.implType {
+	case "openresolv":
+		// Force delete with -f
+		cmd = exec.Command(resolvconfCommand, "-f", "-d", r.ifaceName)
+	default:
+		cmd = exec.Command(resolvconfCommand, "-d", r.ifaceName)
+	}
+
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("delete resolvconf config: %w, output: %s", err, out)
+	}
+
+	return nil
+}
+
+// CleanupUncleanShutdown removes any DNS configuration left over from a previous crash
+// For resolvconf, we attempt to delete any entry for the interface name.
+// This ensures that if the process crashed while DNS was configured, the stale
+// entry is removed on the next startup.
+func (r *ResolvconfDNSConfigurator) CleanupUncleanShutdown() error {
+	// Try to delete any existing entry for this interface
+	// This is idempotent - if no entry exists, resolvconf will just return success
+	var cmd *exec.Cmd
+
+	switch r.implType {
+	case "openresolv":
+		cmd = exec.Command(resolvconfCommand, "-f", "-d", r.ifaceName)
+	default:
+		cmd = exec.Command(resolvconfCommand, "-d", r.ifaceName)
+	}
+
+	// Ignore errors - the entry may not exist, which is fine
+	_ = cmd.Run()
+
+	return nil
+}
+
+// GetCurrentDNS returns the currently configured DNS servers
+func (r *ResolvconfDNSConfigurator) GetCurrentDNS() ([]netip.Addr, error) {
+	// resolvconf doesn't provide a direct way to query per-interface DNS
+	// We can try to read /etc/resolv.conf but it's merged from all sources
+	content, err := exec.Command(resolvconfCommand, "-l").CombinedOutput()
+	if err != nil {
+		// Fall back to reading resolv.conf
+		return readResolvConfServers()
+	}
+
+	// Parse the output (format varies by implementation)
+	return parseResolvconfOutput(string(content)), nil
+}
+
+// applyDNSServers applies DNS server configuration via resolvconf
+func (r *ResolvconfDNSConfigurator) applyDNSServers(servers []netip.Addr) error {
+	if len(servers) == 0 {
+		return fmt.Errorf("no DNS servers provided")
+	}
+
+	// Build resolv.conf content
+	var content bytes.Buffer
+	content.WriteString("# Generated by Olm DNS Manager\n\n")
+
+	for _, server := range servers {
+		content.WriteString("nameserver ")
+		content.WriteString(server.String())
+		content.WriteString("\n")
+	}
+
+	// Apply via resolvconf
+	var cmd *exec.Cmd
+	switch r.implType {
+	case "openresolv":
+		// OpenResolv supports exclusive mode with -x
+		cmd = exec.Command(resolvconfCommand, "-x", "-a", r.ifaceName)
+	default:
+		cmd = exec.Command(resolvconfCommand, "-a", r.ifaceName)
+	}
+
+	cmd.Stdin = &content
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("apply resolvconf config: %w, output: %s", err, out)
+	}
+
+	return nil
+}
+
+// detectResolvconfType detects which resolvconf implementation is being used
+func detectResolvconfType() (string, error) {
+	cmd := exec.Command(resolvconfCommand, "--version")
+	out, err := cmd.CombinedOutput()
+	if err != nil {
+		return "", fmt.Errorf("detect resolvconf type: %w", err)
+	}
+
+	if strings.Contains(string(out), "openresolv") {
+		return "openresolv", nil
+	}
+
+	return "resolvconf", nil
+}
+
+// parseResolvconfOutput parses resolvconf -l output for DNS servers
+func parseResolvconfOutput(output string) []netip.Addr {
+	var servers []netip.Addr
+
+	lines := strings.Split(output, "\n")
+	for _, line := range lines {
+		line = strings.TrimSpace(line)
+
+		// Skip comments and empty lines
+		if line == "" || strings.HasPrefix(line, "#") {
+			continue
+		}
+
+		// Look for nameserver lines
+		if strings.HasPrefix(line, "nameserver") {
+			fields := strings.Fields(line)
+			if len(fields) >= 2 {
+				if addr, err := netip.ParseAddr(fields[1]); err == nil {
+					servers = append(servers, addr)
+				}
+			}
+		}
+	}
+
+	return servers
+}
+
+// readResolvConfServers reads DNS servers from /etc/resolv.conf
+func readResolvConfServers() ([]netip.Addr, error) {
+	cmd := exec.Command("cat", "/etc/resolv.conf")
+	out, err := cmd.CombinedOutput()
+	if err != nil {
+		return nil, fmt.Errorf("read resolv.conf: %w", err)
+	}
+
+	return parseResolvconfOutput(string(out)), nil
+}
+
+// IsResolvconfAvailable checks if resolvconf is available
+func IsResolvconfAvailable() bool {
+	cmd := exec.Command(resolvconfCommand, "--version")
+	return cmd.Run() == nil
+}

dns/platform/systemd.go

@@ -0,0 +1,304 @@
+//go:build linux && !android
+
+package dns
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"net/netip"
+	"time"
+
+	dbus "github.com/godbus/dbus/v5"
+	"golang.org/x/sys/unix"
+)
+
+const (
+	systemdResolvedDest              = "org.freedesktop.resolve1"
+	systemdDbusObjectNode            = "/org/freedesktop/resolve1"
+	systemdDbusManagerIface          = "org.freedesktop.resolve1.Manager"
+	systemdDbusGetLinkMethod         = systemdDbusManagerIface + ".GetLink"
+	systemdDbusFlushCachesMethod     = systemdDbusManagerIface + ".FlushCaches"
+	systemdDbusLinkInterface         = "org.freedesktop.resolve1.Link"
+	systemdDbusSetDNSMethod          = systemdDbusLinkInterface + ".SetDNS"
+	systemdDbusSetDefaultRouteMethod = systemdDbusLinkInterface + ".SetDefaultRoute"
+	systemdDbusSetDomainsMethod      = systemdDbusLinkInterface + ".SetDomains"
+	systemdDbusSetDNSSECMethod       = systemdDbusLinkInterface + ".SetDNSSEC"
+	systemdDbusSetDNSOverTLSMethod   = systemdDbusLinkInterface + ".SetDNSOverTLS"
+	systemdDbusRevertMethod          = systemdDbusLinkInterface + ".Revert"
+
+	// RootZone is the root DNS zone that matches all queries
+	RootZone = "."
+)
+
+// systemdDbusDNSInput maps to (iay) dbus input for SetDNS method
+type systemdDbusDNSInput struct {
+	Family  int32
+	Address []byte
+}
+
+// systemdDbusDomainsInput maps to (sb) dbus input for SetDomains method
+type systemdDbusDomainsInput struct {
+	Domain    string
+	MatchOnly bool
+}
+
+// SystemdResolvedDNSConfigurator manages DNS settings using systemd-resolved D-Bus API
+type SystemdResolvedDNSConfigurator struct {
+	ifaceName      string
+	dbusLinkObject dbus.ObjectPath
+	originalState  *DNSState
+}
+
+// NewSystemdResolvedDNSConfigurator creates a new systemd-resolved DNS configurator
+func NewSystemdResolvedDNSConfigurator(ifaceName string) (*SystemdResolvedDNSConfigurator, error) {
+	// Get network interface
+	iface, err := net.InterfaceByName(ifaceName)
+	if err != nil {
+		return nil, fmt.Errorf("get interface: %w", err)
+	}
+
+	// Connect to D-Bus
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return nil, fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(systemdResolvedDest, systemdDbusObjectNode)
+
+	// Get the link object for this interface
+	var linkPath string
+	if err := obj.Call(systemdDbusGetLinkMethod, 0, iface.Index).Store(&linkPath); err != nil {
+		return nil, fmt.Errorf("get link: %w", err)
+	}
+
+	config := &SystemdResolvedDNSConfigurator{
+		ifaceName:      ifaceName,
+		dbusLinkObject: dbus.ObjectPath(linkPath),
+	}
+
+	// Call cleanup function here
+	if err := config.CleanupUncleanShutdown(); err != nil {
+		fmt.Printf("warning: cleanup unclean shutdown failed: %v\n", err)
+	}
+
+	return config, nil
+}
+
+// Name returns the configurator name
+func (s *SystemdResolvedDNSConfigurator) Name() string {
+	return "systemd-resolved"
+}
+
+// SetDNS sets the DNS servers and returns the original servers
+func (s *SystemdResolvedDNSConfigurator) SetDNS(servers []netip.Addr) ([]netip.Addr, error) {
+	// Get current DNS settings before overriding
+	originalServers, err := s.GetCurrentDNS()
+	if err != nil {
+		// If we can't get current DNS, proceed anyway
+		originalServers = []netip.Addr{}
+	}
+
+	// Store original state
+	s.originalState = &DNSState{
+		OriginalServers:  originalServers,
+		ConfiguratorName: s.Name(),
+	}
+
+	// Apply new DNS servers
+	if err := s.applyDNSServers(servers); err != nil {
+		return nil, fmt.Errorf("apply DNS servers: %w", err)
+	}
+
+	return originalServers, nil
+}
+
+// RestoreDNS restores the original DNS configuration
+func (s *SystemdResolvedDNSConfigurator) RestoreDNS() error {
+	// Call Revert method to restore systemd-resolved defaults
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(systemdResolvedDest, s.dbusLinkObject)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	if err := obj.CallWithContext(ctx, systemdDbusRevertMethod, 0).Store(); err != nil {
+		return fmt.Errorf("revert DNS settings: %w", err)
+	}
+
+	// Flush DNS cache after reverting
+	if err := s.flushDNSCache(); err != nil {
+		fmt.Printf("warning: failed to flush DNS cache: %v\n", err)
+	}
+
+	return nil
+}
+
+// CleanupUncleanShutdown removes any DNS configuration left over from a previous crash
+// For systemd-resolved, the DNS configuration is tied to the network interface.
+// When the interface is destroyed and recreated, systemd-resolved automatically
+// clears the per-link DNS settings, so there's nothing to clean up.
+func (s *SystemdResolvedDNSConfigurator) CleanupUncleanShutdown() error {
+	// systemd-resolved DNS configuration is per-link and automatically cleared
+	// when the link (interface) is destroyed. Since the WireGuard interface is
+	// recreated on restart, there's no leftover state to clean up.
+	return nil
+}
+
+// GetCurrentDNS returns the currently configured DNS servers
+// Note: systemd-resolved doesn't easily expose current per-link DNS servers via D-Bus
+// This is a placeholder that returns an empty list
+func (s *SystemdResolvedDNSConfigurator) GetCurrentDNS() ([]netip.Addr, error) {
+	// systemd-resolved's D-Bus API doesn't have a simple way to query current DNS servers
+	// We would need to parse resolvectl status output or read from /run/systemd/resolve/
+	// For now, return empty list
+	return []netip.Addr{}, nil
+}
+
+// applyDNSServers applies DNS server configuration via systemd-resolved
+func (s *SystemdResolvedDNSConfigurator) applyDNSServers(servers []netip.Addr) error {
+	if len(servers) == 0 {
+		return fmt.Errorf("no DNS servers provided")
+	}
+
+	// Convert servers to systemd-resolved format
+	var dnsInputs []systemdDbusDNSInput
+	for _, server := range servers {
+		family := unix.AF_INET
+		if server.Is6() {
+			family = unix.AF_INET6
+		}
+
+		dnsInputs = append(dnsInputs, systemdDbusDNSInput{
+			Family:  int32(family),
+			Address: server.AsSlice(),
+		})
+	}
+
+	// Connect to D-Bus
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(systemdResolvedDest, s.dbusLinkObject)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	// Call SetDNS method to set the DNS servers
+	if err := obj.CallWithContext(ctx, systemdDbusSetDNSMethod, 0, dnsInputs).Store(); err != nil {
+		return fmt.Errorf("set DNS servers: %w", err)
+	}
+
+	// Set this interface as the default route for DNS
+	// This ensures all DNS queries prefer this interface
+	if err := s.callLinkMethod(systemdDbusSetDefaultRouteMethod, true); err != nil {
+		return fmt.Errorf("set default route: %w", err)
+	}
+
+	// Set the root zone "." as a match-only domain
+	// This captures ALL DNS queries and routes them through this interface
+	domainsInput := []systemdDbusDomainsInput{
+		{
+			Domain:    RootZone,
+			MatchOnly: true,
+		},
+	}
+	if err := s.callLinkMethod(systemdDbusSetDomainsMethod, domainsInput); err != nil {
+		return fmt.Errorf("set domains: %w", err)
+	}
+
+	// Disable DNSSEC - we don't support it and it may be enabled by default
+	if err := s.callLinkMethod(systemdDbusSetDNSSECMethod, "no"); err != nil {
+		// Log warning but don't fail - this is optional
+		fmt.Printf("warning: failed to disable DNSSEC: %v\n", err)
+	}
+
+	// Disable DNSOverTLS - we don't support it and it may be enabled by default
+	if err := s.callLinkMethod(systemdDbusSetDNSOverTLSMethod, "no"); err != nil {
+		// Log warning but don't fail - this is optional
+		fmt.Printf("warning: failed to disable DNSOverTLS: %v\n", err)
+	}
+
+	// Flush DNS cache to ensure new settings take effect immediately
+	if err := s.flushDNSCache(); err != nil {
+		fmt.Printf("warning: failed to flush DNS cache: %v\n", err)
+	}
+
+	return nil
+}
+
+// callLinkMethod is a helper to call methods on the link object
+func (s *SystemdResolvedDNSConfigurator) callLinkMethod(method string, value any) error {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(systemdResolvedDest, s.dbusLinkObject)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	if value != nil {
+		if err := obj.CallWithContext(ctx, method, 0, value).Store(); err != nil {
+			return fmt.Errorf("call %s: %w", method, err)
+		}
+	} else {
+		if err := obj.CallWithContext(ctx, method, 0).Store(); err != nil {
+			return fmt.Errorf("call %s: %w", method, err)
+		}
+	}
+
+	return nil
+}
+
+// flushDNSCache flushes the systemd-resolved DNS cache
+func (s *SystemdResolvedDNSConfigurator) flushDNSCache() error {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return fmt.Errorf("connect to system bus: %w", err)
+	}
+	defer conn.Close()
+
+	obj := conn.Object(systemdResolvedDest, systemdDbusObjectNode)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	if err := obj.CallWithContext(ctx, systemdDbusFlushCachesMethod, 0).Store(); err != nil {
+		return fmt.Errorf("flush caches: %w", err)
+	}
+
+	return nil
+}
+
+// IsSystemdResolvedAvailable checks if systemd-resolved is available and responsive
+func IsSystemdResolvedAvailable() bool {
+	conn, err := dbus.SystemBus()
+	if err != nil {
+		return false
+	}
+	defer conn.Close()
+
+	obj := conn.Object(systemdResolvedDest, systemdDbusObjectNode)
+
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+
+	// Try to ping systemd-resolved
+	if err := obj.CallWithContext(ctx, "org.freedesktop.DBus.Peer.Ping", 0).Store(); err != nil {
+		return false
+	}
+
+	return true
+}

dns/platform/types.go

@@ -0,0 +1,45 @@
+package dns
+
+import "net/netip"
+
+// DNSConfigurator provides an interface for managing system DNS settings
+// across different platforms and implementations
+type DNSConfigurator interface {
+	// SetDNS overrides the system DNS servers with the specified ones
+	// Returns the original DNS servers that were replaced
+	SetDNS(servers []netip.Addr) ([]netip.Addr, error)
+
+	// RestoreDNS restores the original DNS servers
+	RestoreDNS() error
+
+	// GetCurrentDNS returns the currently configured DNS servers
+	GetCurrentDNS() ([]netip.Addr, error)
+
+	// Name returns the name of this configurator implementation
+	Name() string
+
+	// CleanupUncleanShutdown removes any DNS configuration left over from
+	// a previous crash or unclean shutdown. This should be called on startup.
+	CleanupUncleanShutdown() error
+}
+
+// DNSConfig contains the configuration for DNS override
+type DNSConfig struct {
+	// Servers is the list of DNS servers to use
+	Servers []netip.Addr
+
+	// SearchDomains is an optional list of search domains
+	SearchDomains []string
+}
+
+// DNSState represents the saved state of DNS configuration
+type DNSState struct {
+	// OriginalServers are the DNS servers before override
+	OriginalServers []netip.Addr
+
+	// OriginalSearchDomains are the search domains before override
+	OriginalSearchDomains []string
+
+	// ConfiguratorName is the name of the configurator that saved this state
+	ConfiguratorName string
+}

dns/platform/windows.go

@@ -0,0 +1,355 @@
+//go:build windows
+
+package dns
+
+import (
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"net/netip"
+	"syscall"
+	"unsafe"
+
+	"golang.org/x/sys/windows"
+	"golang.org/x/sys/windows/registry"
+)
+
+var (
+	dnsapi                  = syscall.NewLazyDLL("dnsapi.dll")
+	dnsFlushResolverCacheFn = dnsapi.NewProc("DnsFlushResolverCache")
+)
+
+const (
+	interfaceConfigPath           = `SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\Interfaces`
+	interfaceConfigNameServer     = "NameServer"
+	interfaceConfigDhcpNameServer = "DhcpNameServer"
+)
+
+// WindowsDNSConfigurator manages DNS settings on Windows using the registry
+type WindowsDNSConfigurator struct {
+	guid          string
+	originalState *DNSState
+}
+
+// NewWindowsDNSConfigurator creates a new Windows DNS configurator
+// Accepts an interface name and extracts the GUID internally
+func NewWindowsDNSConfigurator(interfaceName string) (*WindowsDNSConfigurator, error) {
+	if interfaceName == "" {
+		return nil, fmt.Errorf("interface name is required")
+	}
+
+	guid, err := getInterfaceGUIDString(interfaceName)
+	if err != nil {
+		return nil, fmt.Errorf("failed to get interface GUID: %w", err)
+	}
+
+	return &WindowsDNSConfigurator{
+		guid: guid,
+	}, nil
+}
+
+// newWindowsDNSConfiguratorFromGUID creates a configurator from a GUID string
+// This is an internal function for use by DetectBestConfigurator
+func newWindowsDNSConfiguratorFromGUID(guid string) (*WindowsDNSConfigurator, error) {
+	if guid == "" {
+		return nil, fmt.Errorf("interface GUID is required")
+	}
+
+	return &WindowsDNSConfigurator{
+		guid: guid,
+	}, nil
+}
+
+// Name returns the configurator name
+func (w *WindowsDNSConfigurator) Name() string {
+	return "windows-registry"
+}
+
+// SetDNS sets the DNS servers and returns the original servers
+func (w *WindowsDNSConfigurator) SetDNS(servers []netip.Addr) ([]netip.Addr, error) {
+	// Get current DNS settings before overriding
+	originalServers, err := w.GetCurrentDNS()
+	if err != nil {
+		return nil, fmt.Errorf("get current DNS: %w", err)
+	}
+
+	// Store original state
+	w.originalState = &DNSState{
+		OriginalServers:  originalServers,
+		ConfiguratorName: w.Name(),
+	}
+
+	// Set new DNS servers
+	if err := w.setDNSServers(servers); err != nil {
+		return nil, fmt.Errorf("set DNS servers: %w", err)
+	}
+
+	// Flush DNS cache
+	if err := w.flushDNSCache(); err != nil {
+		// Non-fatal, just log
+		fmt.Printf("warning: failed to flush DNS cache: %v\n", err)
+	}
+
+	return originalServers, nil
+}
+
+// RestoreDNS restores the original DNS configuration
+func (w *WindowsDNSConfigurator) RestoreDNS() error {
+	if w.originalState == nil {
+		return fmt.Errorf("no original state to restore")
+	}
+
+	// Clear the static DNS setting
+	if err := w.clearDNSServers(); err != nil {
+		return fmt.Errorf("clear DNS servers: %w", err)
+	}
+
+	// Flush DNS cache
+	if err := w.flushDNSCache(); err != nil {
+		fmt.Printf("warning: failed to flush DNS cache: %v\n", err)
+	}
+
+	return nil
+}
+
+// CleanupUncleanShutdown removes any DNS configuration left over from a previous crash
+// On Windows, we rely on the registry-based approach which doesn't leave orphaned state
+// in the same way as macOS scutil. The DNS settings are tied to the interface which
+// gets recreated on restart.
+func (w *WindowsDNSConfigurator) CleanupUncleanShutdown() error {
+	// Windows DNS configuration via registry is interface-specific.
+	// When the WireGuard interface is recreated, it gets a new GUID,
+	// so there's no leftover state to clean up from previous sessions.
+	// The old interface's registry keys are effectively orphaned but harmless.
+	return nil
+}
+
+// GetCurrentDNS returns the currently configured DNS servers
+func (w *WindowsDNSConfigurator) GetCurrentDNS() ([]netip.Addr, error) {
+	regKey, err := w.getInterfaceRegistryKey(registry.QUERY_VALUE)
+	if err != nil {
+		return nil, fmt.Errorf("get interface registry key: %w", err)
+	}
+	defer closeKey(regKey)
+
+	// Try to get static DNS first
+	nameServer, _, err := regKey.GetStringValue(interfaceConfigNameServer)
+	if err == nil && nameServer != "" {
+		return w.parseServerList(nameServer), nil
+	}
+
+	// Fall back to DHCP DNS
+	dhcpNameServer, _, err := regKey.GetStringValue(interfaceConfigDhcpNameServer)
+	if err == nil && dhcpNameServer != "" {
+		return w.parseServerList(dhcpNameServer), nil
+	}
+
+	return []netip.Addr{}, nil
+}
+
+// setDNSServers sets the DNS servers in the registry
+func (w *WindowsDNSConfigurator) setDNSServers(servers []netip.Addr) error {
+	if len(servers) == 0 {
+		return fmt.Errorf("no DNS servers provided")
+	}
+
+	regKey, err := w.getInterfaceRegistryKey(registry.SET_VALUE)
+	if err != nil {
+		return fmt.Errorf("get interface registry key: %w", err)
+	}
+	defer closeKey(regKey)
+
+	// Build comma-separated or space-separated list of servers
+	var serverList string
+	for i, server := range servers {
+		if i > 0 {
+			serverList += ","
+		}
+		serverList += server.String()
+	}
+
+	if err := regKey.SetStringValue(interfaceConfigNameServer, serverList); err != nil {
+		return fmt.Errorf("set NameServer: %w", err)
+	}
+
+	return nil
+}
+
+// clearDNSServers clears the static DNS server setting
+func (w *WindowsDNSConfigurator) clearDNSServers() error {
+	regKey, err := w.getInterfaceRegistryKey(registry.SET_VALUE)
+	if err != nil {
+		return fmt.Errorf("get interface registry key: %w", err)
+	}
+	defer closeKey(regKey)
+
+	// Set empty string to revert to DHCP
+	if err := regKey.SetStringValue(interfaceConfigNameServer, ""); err != nil {
+		return fmt.Errorf("clear NameServer: %w", err)
+	}
+
+	return nil
+}
+
+// getInterfaceRegistryKey opens the registry key for the network interface
+func (w *WindowsDNSConfigurator) getInterfaceRegistryKey(access uint32) (registry.Key, error) {
+	regKeyPath := interfaceConfigPath + `\` + w.guid
+
+	regKey, err := registry.OpenKey(registry.LOCAL_MACHINE, regKeyPath, access)
+	if err != nil {
+		return 0, fmt.Errorf("open HKEY_LOCAL_MACHINE\\%s: %w", regKeyPath, err)
+	}
+
+	return regKey, nil
+}
+
+// parseServerList parses a comma or space-separated list of DNS servers
+func (w *WindowsDNSConfigurator) parseServerList(serverList string) []netip.Addr {
+	var servers []netip.Addr
+
+	// Split by comma or space
+	parts := splitByDelimiters(serverList, []rune{',', ' '})
+
+	for _, part := range parts {
+		if addr, err := netip.ParseAddr(part); err == nil {
+			servers = append(servers, addr)
+		}
+	}
+
+	return servers
+}
+
+// flushDNSCache flushes the Windows DNS resolver cache
+func (w *WindowsDNSConfigurator) flushDNSCache() error {
+	// dnsFlushResolverCacheFn.Call() may panic if the func is not found
+	defer func() {
+		if rec := recover(); rec != nil {
+			fmt.Printf("warning: DnsFlushResolverCache panicked: %v\n", rec)
+		}
+	}()
+
+	ret, _, err := dnsFlushResolverCacheFn.Call()
+	if ret == 0 {
+		if err != nil && !errors.Is(err, syscall.Errno(0)) {
+			return fmt.Errorf("DnsFlushResolverCache failed: %w", err)
+		}
+		return fmt.Errorf("DnsFlushResolverCache failed")
+	}
+
+	return nil
+}
+
+// splitByDelimiters splits a string by multiple delimiters
+func splitByDelimiters(s string, delimiters []rune) []string {
+	var result []string
+	var current []rune
+
+	for _, char := range s {
+		isDelimiter := false
+		for _, delim := range delimiters {
+			if char == delim {
+				isDelimiter = true
+				break
+			}
+		}
+
+		if isDelimiter {
+			if len(current) > 0 {
+				result = append(result, string(current))
+				current = []rune{}
+			}
+		} else {
+			current = append(current, char)
+		}
+	}
+
+	if len(current) > 0 {
+		result = append(result, string(current))
+	}
+
+	return result
+}
+
+// closeKey closes a registry key and logs errors
+func closeKey(closer io.Closer) {
+	if err := closer.Close(); err != nil {
+		fmt.Printf("warning: failed to close registry key: %v\n", err)
+	}
+}
+
+// getInterfaceGUIDString retrieves the GUID string for a Windows TUN interface
+// This is required for registry-based DNS configuration on Windows
+func getInterfaceGUIDString(interfaceName string) (string, error) {
+	if interfaceName == "" {
+		return "", fmt.Errorf("interface name is required")
+	}
+
+	iface, err := net.InterfaceByName(interfaceName)
+	if err != nil {
+		return "", fmt.Errorf("failed to get interface %s: %w", interfaceName, err)
+	}
+
+	luid, err := indexToLUID(uint32(iface.Index))
+	if err != nil {
+		return "", fmt.Errorf("failed to convert index to LUID: %w", err)
+	}
+
+	// Convert LUID to GUID using Windows API
+	guid, err := luidToGUID(luid)
+	if err != nil {
+		return "", fmt.Errorf("failed to convert LUID to GUID: %w", err)
+	}
+
+	return guid, nil
+}
+
+// indexToLUID converts a Windows interface index to a LUID
+func indexToLUID(index uint32) (uint64, error) {
+	var luid uint64
+
+	// Load the iphlpapi.dll and get the ConvertInterfaceIndexToLuid function
+	iphlpapi := windows.NewLazySystemDLL("iphlpapi.dll")
+	convertInterfaceIndexToLuid := iphlpapi.NewProc("ConvertInterfaceIndexToLuid")
+
+	// Call the Windows API
+	ret, _, err := convertInterfaceIndexToLuid.Call(
+		uintptr(index),
+		uintptr(unsafe.Pointer(&luid)),
+	)
+
+	if ret != 0 {
+		return 0, fmt.Errorf("ConvertInterfaceIndexToLuid failed with code %d: %w", ret, err)
+	}
+
+	return luid, nil
+}
+
+// luidToGUID converts a Windows LUID (Locally Unique Identifier) to a GUID string
+// using the Windows ConvertInterface* APIs
+func luidToGUID(luid uint64) (string, error) {
+	var guid windows.GUID
+
+	// Load the iphlpapi.dll and get the ConvertInterfaceLuidToGuid function
+	iphlpapi := windows.NewLazySystemDLL("iphlpapi.dll")
+	convertLuidToGuid := iphlpapi.NewProc("ConvertInterfaceLuidToGuid")
+
+	// Call the Windows API
+	// NET_LUID is a 64-bit value on Windows
+	ret, _, err := convertLuidToGuid.Call(
+		uintptr(unsafe.Pointer(&luid)),
+		uintptr(unsafe.Pointer(&guid)),
+	)
+
+	if ret != 0 {
+		return "", fmt.Errorf("ConvertInterfaceLuidToGuid failed with code %d: %w", ret, err)
+	}
+
+	// Format the GUID as a string with curly braces
+	guidStr := fmt.Sprintf("{%08X-%04X-%04X-%02X%02X-%02X%02X%02X%02X%02X%02X}",
+		guid.Data1, guid.Data2, guid.Data3,
+		guid.Data4[0], guid.Data4[1], guid.Data4[2], guid.Data4[3],
+		guid.Data4[4], guid.Data4[5], guid.Data4[6], guid.Data4[7])
+
+	return guidStr, nil
+}

docker-compose.yml

@@ -1,9 +1,15 @@
 services:
-  newt:
-    image: fosrl/newt:latest
-    container_name: newt
+  olm:
+    image: fosrl/olm:latest
+    container_name: olm
     restart: unless-stopped
     environment:
       - PANGOLIN_ENDPOINT=https://example.com
-      - NEWT_ID=2ix2t8xk22ubpfy 
-      - NEWT_SECRET=nnisrfsdfc7prqsp9ewo1dvtvci50j5uiqotez00dgap0ii2 
+      - OLM_ID=vdqnz8rwgb95cnp 
+      - OLM_SECRET=1sw05qv1tkfdb1k81zpw05nahnnjvmhxjvf746umwagddmdg 
+    cap_add:
+      - NET_ADMIN
+      - SYS_MODULE
+    devices:
+      - /dev/net/tun:/dev/net/tun
+    network_mode: host

entrypoint.sh

@@ -4,7 +4,7 @@ set -e
 
 # first arg is `-f` or `--some-option`
 if [ "${1#-}" != "$1" ]; then
-    set -- newt "$@"
+    set -- olm "$@"
 fi
 
 exec "$@"

get-olm.sh

@@ -0,0 +1,279 @@
+#!/bin/bash
+
+# Get Olm - Cross-platform installation script
+# Usage: curl -fsSL https://raw.githubusercontent.com/fosrl/olm/refs/heads/main/get-olm.sh | bash
+
+set -e
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+# GitHub repository info
+REPO="fosrl/olm"
+GITHUB_API_URL="https://api.github.com/repos/${REPO}/releases/latest"
+
+# Function to print colored output
+print_status() {
+    echo -e "${GREEN}[INFO]${NC} $1"
+}
+
+print_warning() {
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+print_error() {
+    echo -e "${RED}[ERROR]${NC} $1"
+}
+
+# Function to get latest version from GitHub API
+get_latest_version() {
+    local latest_info
+    
+    if command -v curl >/dev/null 2>&1; then
+        latest_info=$(curl -fsSL "$GITHUB_API_URL" 2>/dev/null)
+    elif command -v wget >/dev/null 2>&1; then
+        latest_info=$(wget -qO- "$GITHUB_API_URL" 2>/dev/null)
+    else
+        print_error "Neither curl nor wget is available. Please install one of them." >&2
+        exit 1
+    fi
+    
+    if [ -z "$latest_info" ]; then
+        print_error "Failed to fetch latest version information" >&2
+        exit 1
+    fi
+    
+    # Extract version from JSON response (works without jq)
+    local version=$(echo "$latest_info" | grep '"tag_name"' | head -1 | sed 's/.*"tag_name": *"\([^"]*\)".*/\1/')
+    
+    if [ -z "$version" ]; then
+        print_error "Could not parse version from GitHub API response" >&2
+        exit 1
+    fi
+    
+    # Remove 'v' prefix if present
+    version=$(echo "$version" | sed 's/^v//')
+    
+    echo "$version"
+}
+
+# Detect OS and architecture
+detect_platform() {
+    local os arch
+    
+    # Detect OS
+    case "$(uname -s)" in
+        Linux*)     os="linux" ;;
+        Darwin*)    os="darwin" ;;
+        MINGW*|MSYS*|CYGWIN*) os="windows" ;;
+        FreeBSD*)   os="freebsd" ;;
+        *)
+            print_error "Unsupported operating system: $(uname -s)"
+            exit 1
+            ;;
+    esac
+    
+    # Detect architecture
+    case "$(uname -m)" in
+        x86_64|amd64)   arch="amd64" ;;
+        arm64|aarch64)  arch="arm64" ;;
+        armv7l|armv6l)  
+            if [ "$os" = "linux" ]; then
+                if [ "$(uname -m)" = "armv6l" ]; then
+                    arch="arm32v6"
+                else
+                    arch="arm32"
+                fi
+            else
+                arch="arm64"  # Default for non-Linux ARM
+            fi
+            ;;
+        riscv64)        
+            if [ "$os" = "linux" ]; then
+                arch="riscv64"
+            else
+                print_error "RISC-V architecture only supported on Linux"
+                exit 1
+            fi
+            ;;
+        *)
+            print_error "Unsupported architecture: $(uname -m)"
+            exit 1
+            ;;
+    esac
+    
+    echo "${os}_${arch}"
+}
+
+# Get installation directory
+get_install_dir() {
+    local platform="$1"
+    
+    if [[ "$platform" == *"windows"* ]]; then
+        echo "$HOME/bin"
+    else
+        # For Unix-like systems, prioritize system-wide directories for sudo access
+        # Check in order of preference: /usr/local/bin, /usr/bin, ~/.local/bin
+        if [ -d "/usr/local/bin" ]; then
+            echo "/usr/local/bin"
+        elif [ -d "/usr/bin" ]; then
+            echo "/usr/bin"
+        else
+            # Fallback to user directory if system directories don't exist
+            echo "$HOME/.local/bin"
+        fi
+    fi
+}
+
+# Check if we need sudo for installation
+need_sudo() {
+    local install_dir="$1"
+    
+    # If installing to system directory and we don't have write permission, need sudo
+    if [[ "$install_dir" == "/usr/local/bin" || "$install_dir" == "/usr/bin" ]]; then
+        if [ ! -w "$install_dir" ] 2>/dev/null; then
+            return 0  # Need sudo
+        fi
+    fi
+    return 1  # Don't need sudo
+}
+
+# Download and install olm
+install_olm() {
+    local platform="$1"
+    local install_dir="$2"
+    local binary_name="olm_${platform}"
+    local exe_suffix=""
+    
+    # Add .exe suffix for Windows
+    if [[ "$platform" == *"windows"* ]]; then
+        binary_name="${binary_name}.exe"
+        exe_suffix=".exe"
+    fi
+    
+    local download_url="${BASE_URL}/${binary_name}"
+    local temp_file="/tmp/olm${exe_suffix}"
+    local final_path="${install_dir}/olm${exe_suffix}"
+    
+    print_status "Downloading olm from ${download_url}"
+    
+    # Download the binary
+    if command -v curl >/dev/null 2>&1; then
+        curl -fsSL "$download_url" -o "$temp_file"
+    elif command -v wget >/dev/null 2>&1; then
+        wget -q "$download_url" -O "$temp_file"
+    else
+        print_error "Neither curl nor wget is available. Please install one of them."
+        exit 1
+    fi
+    
+    # Check if we need sudo for installation
+    local use_sudo=""
+    if need_sudo "$install_dir"; then
+        print_status "Administrator privileges required for system-wide installation"
+        if command -v sudo >/dev/null 2>&1; then
+            use_sudo="sudo"
+        else
+            print_error "sudo is required for system-wide installation but not available"
+            exit 1
+        fi
+    fi
+    
+    # Create install directory if it doesn't exist
+    if [ -n "$use_sudo" ]; then
+        $use_sudo mkdir -p "$install_dir"
+    else
+        mkdir -p "$install_dir"
+    fi
+    
+    # Move binary to install directory
+    if [ -n "$use_sudo" ]; then
+        $use_sudo mv "$temp_file" "$final_path"
+        $use_sudo chmod +x "$final_path"
+    else
+        mv "$temp_file" "$final_path"
+        chmod +x "$final_path"
+    fi
+    
+    print_status "olm installed to ${final_path}"
+    
+    # Check if install directory is in PATH (only warn for non-system directories)
+    if [[ "$install_dir" != "/usr/local/bin" && "$install_dir" != "/usr/bin" ]]; then
+        if ! echo "$PATH" | grep -q "$install_dir"; then
+            print_warning "Install directory ${install_dir} is not in your PATH."
+            print_warning "Add it to your PATH by adding this line to your shell profile:"
+            print_warning "  export PATH=\"${install_dir}:\$PATH\""
+        fi
+    fi
+}
+
+# Verify installation
+verify_installation() {
+    local install_dir="$1"
+    local exe_suffix=""
+    
+    if [[ "$PLATFORM" == *"windows"* ]]; then
+        exe_suffix=".exe"
+    fi
+    
+    local olm_path="${install_dir}/olm${exe_suffix}"
+    
+    if [ -f "$olm_path" ] && [ -x "$olm_path" ]; then
+        print_status "Installation successful!"
+        print_status "olm version: $("$olm_path" --version 2>/dev/null || echo "unknown")"
+        return 0
+    else
+        print_error "Installation failed. Binary not found or not executable."
+        return 1
+    fi
+}
+
+# Main installation process
+main() {
+    print_status "Installing latest version of olm..."
+    
+    # Get latest version
+    print_status "Fetching latest version from GitHub..."
+    VERSION=$(get_latest_version)
+    print_status "Latest version: v${VERSION}"
+    
+    # Set base URL with the fetched version
+    BASE_URL="https://github.com/${REPO}/releases/download/${VERSION}"
+    
+    # Detect platform
+    PLATFORM=$(detect_platform)
+    print_status "Detected platform: ${PLATFORM}"
+    
+    # Get install directory
+    INSTALL_DIR=$(get_install_dir "$PLATFORM")
+    print_status "Install directory: ${INSTALL_DIR}"
+    
+    # Inform user about system-wide installation
+    if [[ "$INSTALL_DIR" == "/usr/local/bin" || "$INSTALL_DIR" == "/usr/bin" ]]; then
+        print_status "Installing system-wide for sudo access"
+    fi
+    
+    # Install olm
+    install_olm "$PLATFORM" "$INSTALL_DIR"
+    
+    # Verify installation
+    if verify_installation "$INSTALL_DIR"; then
+        print_status "olm is ready to use!"
+        if [[ "$INSTALL_DIR" == "/usr/local/bin" || "$INSTALL_DIR" == "/usr/bin" ]]; then
+            print_status "olm is installed system-wide and accessible via sudo"
+        fi
+        if [[ "$PLATFORM" == *"windows"* ]]; then
+            print_status "Run 'olm --help' to get started"
+        else
+            print_status "Run 'olm --help' or 'sudo olm --help' to get started"
+        fi
+    else
+        exit 1
+    fi
+}
+
+# Run main function
+main "$@"

go.mod

@@ -1,19 +1,32 @@
-module github.com/fosrl/newt
+module github.com/fosrl/olm
 
-go 1.23.1
-
-toolchain go1.23.2
-
-require golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173
+go 1.25
 
 require (
-	github.com/google/btree v1.1.2 // indirect
-	github.com/gorilla/websocket v1.5.3 // indirect
-	golang.org/x/crypto v0.28.0 // indirect
-	golang.org/x/net v0.30.0 // indirect
-	golang.org/x/sys v0.26.0 // indirect
-	golang.org/x/time v0.7.0 // indirect
-	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
-	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6 // indirect
-	gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259 // indirect
+	github.com/Microsoft/go-winio v0.6.2
+	github.com/fosrl/newt v1.8.1
+	github.com/godbus/dbus/v5 v5.2.2
+	github.com/gorilla/websocket v1.5.3
+	github.com/miekg/dns v1.1.70
+	golang.org/x/sys v0.40.0
+	golang.zx2c4.com/wireguard v0.0.0-20250521234502-f333402bd9cb
+	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10
+	gvisor.dev/gvisor v0.0.0-20250503011706-39ed1f5ac29c
+	software.sslmate.com/src/go-pkcs12 v0.7.0
+)
+
+require (
+	github.com/google/btree v1.1.3 // indirect
+	github.com/google/go-cmp v0.7.0 // indirect
+	github.com/vishvananda/netlink v1.3.1 // indirect
+	github.com/vishvananda/netns v0.0.5 // indirect
+	golang.org/x/crypto v0.46.0 // indirect
+	golang.org/x/exp v0.0.0-20251113190631-e25ba8c21ef6 // indirect
+	golang.org/x/mod v0.31.0 // indirect
+	golang.org/x/net v0.48.0 // indirect
+	golang.org/x/sync v0.19.0 // indirect
+	golang.org/x/time v0.12.0 // indirect
+	golang.org/x/tools v0.40.0 // indirect
+	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
+	golang.zx2c4.com/wireguard/windows v0.5.3 // indirect
 )

go.sum

@@ -1,20 +1,48 @@
-github.com/google/btree v1.1.2 h1:xf4v41cLI2Z6FxbKm+8Bu+m8ifhj15JuZ9sa0jZCMUU=
-github.com/google/btree v1.1.2/go.mod h1:qOPhT0dTNdNzV6Z/lhRX0YXUafgPLFUh+gZMl761Gm4=
+github.com/Microsoft/go-winio v0.6.2 h1:F2VQgta7ecxGYO8k3ZZz3RS8fVIXVxONVUPlNERoyfY=
+github.com/Microsoft/go-winio v0.6.2/go.mod h1:yd8OoFMLzJbo9gZq8j5qaps8bJ9aShtEA8Ipt1oGCvU=
+github.com/fosrl/newt v1.8.1 h1:oP3xBEISoO/TENsHccqqs6LXpoOWCt6aiP75CfIWpvk=
+github.com/fosrl/newt v1.8.1/go.mod h1:pol958CEs0nQmo/35Ltv0CGksheIKCS2hoNvdTVLEcI=
+github.com/godbus/dbus/v5 v5.2.2 h1:TUR3TgtSVDmjiXOgAAyaZbYmIeP3DPkld3jgKGV8mXQ=
+github.com/godbus/dbus/v5 v5.2.2/go.mod h1:3AAv2+hPq5rdnr5txxxRwiGjPXamgoIHgz9FPBfOp3c=
+github.com/google/btree v1.1.3 h1:CVpQJjYgC4VbzxeGVHfvZrv1ctoYCAI8vbl07Fcxlyg=
+github.com/google/btree v1.1.3/go.mod h1:qOPhT0dTNdNzV6Z/lhRX0YXUafgPLFUh+gZMl761Gm4=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
 github.com/gorilla/websocket v1.5.3 h1:saDtZ6Pbx/0u+bgYQ3q96pZgCzfhKXGPqt7kZ72aNNg=
 github.com/gorilla/websocket v1.5.3/go.mod h1:YR8l580nyteQvAITg2hZ9XVh4b55+EU/adAjf1fMHhE=
-golang.org/x/crypto v0.28.0 h1:GBDwsMXVQi34v5CCYUm2jkJvu4cbtru2U4TN2PSyQnw=
-golang.org/x/crypto v0.28.0/go.mod h1:rmgy+3RHxRZMyY0jjAJShp2zgEdOqj2AO7U0pYmeQ7U=
-golang.org/x/net v0.30.0 h1:AcW1SDZMkb8IpzCdQUaIq2sP4sZ4zw+55h6ynffypl4=
-golang.org/x/net v0.30.0/go.mod h1:2wGyMJ5iFasEhkwi13ChkO/t1ECNC4X4eBKkVFyYFlU=
-golang.org/x/sys v0.26.0 h1:KHjCJyddX0LoSTb3J+vWpupP9p0oznkqVk/IfjymZbo=
-golang.org/x/sys v0.26.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/time v0.7.0 h1:ntUhktv3OPE6TgYxXWv9vKvUSJyIFJlyohwbkEwPrKQ=
-golang.org/x/time v0.7.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
+github.com/miekg/dns v1.1.70 h1:DZ4u2AV35VJxdD9Fo9fIWm119BsQL5cZU1cQ9s0LkqA=
+github.com/miekg/dns v1.1.70/go.mod h1:+EuEPhdHOsfk6Wk5TT2CzssZdqkmFhf8r+aVyDEToIs=
+github.com/vishvananda/netlink v1.3.1 h1:3AEMt62VKqz90r0tmNhog0r/PpWKmrEShJU0wJW6bV0=
+github.com/vishvananda/netlink v1.3.1/go.mod h1:ARtKouGSTGchR8aMwmkzC0qiNPrrWO5JS/XMVl45+b4=
+github.com/vishvananda/netns v0.0.5 h1:DfiHV+j8bA32MFM7bfEunvT8IAqQ/NzSJHtcmW5zdEY=
+github.com/vishvananda/netns v0.0.5/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
+golang.org/x/crypto v0.46.0 h1:cKRW/pmt1pKAfetfu+RCEvjvZkA9RimPbh7bhFjGVBU=
+golang.org/x/crypto v0.46.0/go.mod h1:Evb/oLKmMraqjZ2iQTwDwvCtJkczlDuTmdJXoZVzqU0=
+golang.org/x/exp v0.0.0-20251113190631-e25ba8c21ef6 h1:zfMcR1Cs4KNuomFFgGefv5N0czO2XZpUbxGUy8i8ug0=
+golang.org/x/exp v0.0.0-20251113190631-e25ba8c21ef6/go.mod h1:46edojNIoXTNOhySWIWdix628clX9ODXwPsQuG6hsK0=
+golang.org/x/mod v0.31.0 h1:HaW9xtz0+kOcWKwli0ZXy79Ix+UW/vOfmWI5QVd2tgI=
+golang.org/x/mod v0.31.0/go.mod h1:43JraMp9cGx1Rx3AqioxrbrhNsLl2l/iNAvuBkrezpg=
+golang.org/x/net v0.48.0 h1:zyQRTTrjc33Lhh0fBgT/H3oZq9WuvRR5gPC70xpDiQU=
+golang.org/x/net v0.48.0/go.mod h1:+ndRgGjkh8FGtu1w1FGbEC31if4VrNVMuKTgcAAnQRY=
+golang.org/x/sync v0.19.0 h1:vV+1eWNmZ5geRlYjzm2adRgW2/mcpevXNg50YZtPCE4=
+golang.org/x/sync v0.19.0/go.mod h1:9KTHXmSnoGruLpwFjVSX0lNNA75CykiMECbovNTZqGI=
+golang.org/x/sys v0.2.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.40.0 h1:DBZZqJ2Rkml6QMQsZywtnjnnGvHza6BTfYFWY9kjEWQ=
+golang.org/x/sys v0.40.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
+golang.org/x/time v0.12.0 h1:ScB/8o8olJvc+CQPWrK3fPZNfh7qgwCrY0zJmoEQLSE=
+golang.org/x/time v0.12.0/go.mod h1:CDIdPxbZBQxdj6cxyCIdrNogrJKMJ7pr37NYpMcMDSg=
+golang.org/x/tools v0.40.0 h1:yLkxfA+Qnul4cs9QA3KnlFu0lVmd8JJfoq+E41uSutA=
+golang.org/x/tools v0.40.0/go.mod h1:Ik/tzLRlbscWpqqMRjyWYDisX8bG13FrdXp3o4Sr9lc=
 golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 h1:B82qJJgjvYKsXS9jeunTOisW56dUokqW/FOteYJJ/yg=
 golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2/go.mod h1:deeaetjYA+DHMHg+sMSMI58GrEteJUUzzw7en6TJQcI=
-golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173 h1:/jFs0duh4rdb8uIfPMv78iAJGcPKDeqAFnaLBropIC4=
-golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173/go.mod h1:tkCQ4FQXmpAgYVh++1cq16/dH4QJtmvpRv19DWGAHSA=
-golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6 h1:CawjfCvYQH2OU3/TnxLx97WDSUDRABfT18pCOYwc2GE=
-golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6/go.mod h1:3rxYc4HtVcSG9gVaTs2GEBdehh+sYPOwKtyUWEOTb80=
-gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259 h1:TbRPT0HtzFP3Cno1zZo7yPzEEnfu8EjLfl6IU9VfqkQ=
-gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259/go.mod h1:AVgIgHMwK63XvmAzWG9vLQ41YnVHN0du0tEC46fI7yY=
+golang.zx2c4.com/wireguard v0.0.0-20250521234502-f333402bd9cb h1:whnFRlWMcXI9d+ZbWg+4sHnLp52d5yiIPUxMBSt4X9A=
+golang.zx2c4.com/wireguard v0.0.0-20250521234502-f333402bd9cb/go.mod h1:rpwXGsirqLqN2L0JDJQlwOboGHmptD5ZD6T2VmcqhTw=
+golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10 h1:3GDAcqdIg1ozBNLgPy4SLT84nfcBjr6rhGtXYtrkWLU=
+golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10/go.mod h1:T97yPqesLiNrOYxkwmhMI0ZIlJDm+p0PMR8eRVeR5tQ=
+golang.zx2c4.com/wireguard/windows v0.5.3 h1:On6j2Rpn3OEMXqBq00QEDC7bWSZrPIHKIus8eIuExIE=
+golang.zx2c4.com/wireguard/windows v0.5.3/go.mod h1:9TEe8TJmtwyQebdFwAkEWOPr3prrtqm+REGFifP60hI=
+gvisor.dev/gvisor v0.0.0-20250503011706-39ed1f5ac29c h1:m/r7OM+Y2Ty1sgBQ7Qb27VgIMBW8ZZhT4gLnUyDIhzI=
+gvisor.dev/gvisor v0.0.0-20250503011706-39ed1f5ac29c/go.mod h1:3r5CMtNQMKIvBlrmM9xWUNamjKBYPOWyXOjmg5Kts3g=
+software.sslmate.com/src/go-pkcs12 v0.7.0 h1:Db8W44cB54TWD7stUFFSWxdfpdn6fZVcDl0w3R4RVM0=
+software.sslmate.com/src/go-pkcs12 v0.7.0/go.mod h1:Qiz0EyvDRJjjxGyUQa2cCNZn/wMyzrRJ/qcDXOQazLI=

logger/level.go

@@ -1,27 +0,0 @@
-package logger
-
-type LogLevel int
-
-const (
-	DEBUG LogLevel = iota
-	INFO
-	WARN
-	ERROR
-	FATAL
-)
-
-var levelStrings = map[LogLevel]string{
-	DEBUG: "DEBUG",
-	INFO:  "INFO",
-	WARN:  "WARN",
-	ERROR: "ERROR",
-	FATAL: "FATAL",
-}
-
-// String returns the string representation of the log level
-func (l LogLevel) String() string {
-	if s, ok := levelStrings[l]; ok {
-		return s
-	}
-	return "UNKNOWN"
-}

logger/logger.go

@@ -1,106 +0,0 @@
-package logger
-
-import (
-	"fmt"
-	"log"
-	"os"
-	"sync"
-	"time"
-)
-
-// Logger struct holds the logger instance
-type Logger struct {
-	logger *log.Logger
-	level  LogLevel
-}
-
-var (
-	defaultLogger *Logger
-	once          sync.Once
-)
-
-// NewLogger creates a new logger instance
-func NewLogger() *Logger {
-	return &Logger{
-		logger: log.New(os.Stdout, "", 0),
-		level:  DEBUG,
-	}
-}
-
-// Init initializes the default logger
-func Init() *Logger {
-	once.Do(func() {
-		defaultLogger = NewLogger()
-	})
-	return defaultLogger
-}
-
-// GetLogger returns the default logger instance
-func GetLogger() *Logger {
-	if defaultLogger == nil {
-		Init()
-	}
-	return defaultLogger
-}
-
-// SetLevel sets the minimum logging level
-func (l *Logger) SetLevel(level LogLevel) {
-	l.level = level
-}
-
-// log handles the actual logging
-func (l *Logger) log(level LogLevel, format string, args ...interface{}) {
-	if level < l.level {
-		return
-	}
-	timestamp := time.Now().Format("2006/01/02 15:04:05")
-	message := fmt.Sprintf(format, args...)
-	l.logger.Printf("%s: %s %s", level.String(), timestamp, message)
-}
-
-// Debug logs debug level messages
-func (l *Logger) Debug(format string, args ...interface{}) {
-	l.log(DEBUG, format, args...)
-}
-
-// Info logs info level messages
-func (l *Logger) Info(format string, args ...interface{}) {
-	l.log(INFO, format, args...)
-}
-
-// Warn logs warning level messages
-func (l *Logger) Warn(format string, args ...interface{}) {
-	l.log(WARN, format, args...)
-}
-
-// Error logs error level messages
-func (l *Logger) Error(format string, args ...interface{}) {
-	l.log(ERROR, format, args...)
-}
-
-// Fatal logs fatal level messages and exits
-func (l *Logger) Fatal(format string, args ...interface{}) {
-	l.log(FATAL, format, args...)
-	os.Exit(1)
-}
-
-// Global helper functions
-func Debug(format string, args ...interface{}) {
-	GetLogger().Debug(format, args...)
-}
-
-func Info(format string, args ...interface{}) {
-	GetLogger().Info(format, args...)
-}
-
-func Warn(format string, args ...interface{}) {
-	GetLogger().Warn(format, args...)
-}
-
-func Error(format string, args ...interface{}) {
-	GetLogger().Error(format, args...)
-}
-
-func Fatal(format string, args ...interface{}) {
-	GetLogger().Fatal(format, args...)
-}

main.go

@@ -1,603 +1,264 @@
 package main
 
 import (
-	"bytes"
-	"encoding/base64"
-	"encoding/hex"
-	"encoding/json"
-	"flag"
+	"context"
 	"fmt"
-	"math/rand"
-	"net"
-	"net/netip"
 	"os"
 	"os/signal"
-	"strings"
+	"runtime"
 	"syscall"
-	"time"
 
 	"github.com/fosrl/newt/logger"
-	"github.com/fosrl/newt/proxy"
-	"github.com/fosrl/newt/websocket"
-
-	"golang.org/x/net/icmp"
-	"golang.org/x/net/ipv4"
-	"golang.zx2c4.com/wireguard/conn"
-	"golang.zx2c4.com/wireguard/device"
-	"golang.zx2c4.com/wireguard/tun"
-	"golang.zx2c4.com/wireguard/tun/netstack"
-	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+	"github.com/fosrl/newt/updates"
+	"github.com/fosrl/olm/olm"
 )
 
-type WgData struct {
-	Endpoint  string        `json:"endpoint"`
-	PublicKey string        `json:"publicKey"`
-	ServerIP  string        `json:"serverIP"`
-	TunnelIP  string        `json:"tunnelIP"`
-	Targets   TargetsByType `json:"targets"`
-}
-
-type TargetsByType struct {
-	UDP []string `json:"udp"`
-	TCP []string `json:"tcp"`
-}
-
-type TargetData struct {
-	Targets []string `json:"targets"`
-}
-
-func fixKey(key string) string {
-	// Remove any whitespace
-	key = strings.TrimSpace(key)
-
-	// Decode from base64
-	decoded, err := base64.StdEncoding.DecodeString(key)
-	if err != nil {
-		logger.Fatal("Error decoding base64:", err)
-	}
-
-	// Convert to hex
-	return hex.EncodeToString(decoded)
-}
-
-func ping(tnet *netstack.Net, dst string) error {
-	logger.Info("Pinging %s", dst)
-	socket, err := tnet.Dial("ping4", dst)
-	if err != nil {
-		return fmt.Errorf("failed to create ICMP socket: %w", err)
-	}
-	defer socket.Close()
-
-	requestPing := icmp.Echo{
-		Seq:  rand.Intn(1 << 16),
-		Data: []byte("gopher burrow"),
-	}
-
-	icmpBytes, err := (&icmp.Message{Type: ipv4.ICMPTypeEcho, Code: 0, Body: &requestPing}).Marshal(nil)
-	if err != nil {
-		return fmt.Errorf("failed to marshal ICMP message: %w", err)
-	}
-
-	if err := socket.SetReadDeadline(time.Now().Add(time.Second * 10)); err != nil {
-		return fmt.Errorf("failed to set read deadline: %w", err)
-	}
-
-	start := time.Now()
-	_, err = socket.Write(icmpBytes)
-	if err != nil {
-		return fmt.Errorf("failed to write ICMP packet: %w", err)
-	}
-
-	n, err := socket.Read(icmpBytes[:])
-	if err != nil {
-		return fmt.Errorf("failed to read ICMP packet: %w", err)
-	}
-
-	replyPacket, err := icmp.ParseMessage(1, icmpBytes[:n])
-	if err != nil {
-		return fmt.Errorf("failed to parse ICMP packet: %w", err)
-	}
-
-	replyPing, ok := replyPacket.Body.(*icmp.Echo)
-	if !ok {
-		return fmt.Errorf("invalid reply type: got %T, want *icmp.Echo", replyPacket.Body)
-	}
-
-	if !bytes.Equal(replyPing.Data, requestPing.Data) || replyPing.Seq != requestPing.Seq {
-		return fmt.Errorf("invalid ping reply: got seq=%d data=%q, want seq=%d data=%q",
-			replyPing.Seq, replyPing.Data, requestPing.Seq, requestPing.Data)
-	}
-
-	logger.Info("Ping latency: %v", time.Since(start))
-	return nil
-}
-
-func startPingCheck(tnet *netstack.Net, serverIP string, stopChan chan struct{}) {
-	ticker := time.NewTicker(10 * time.Second)
-	defer ticker.Stop()
-
-	go func() {
-		for {
-			select {
-			case <-ticker.C:
-				err := ping(tnet, serverIP)
-				if err != nil {
-					logger.Warn("Periodic ping failed: %v", err)
-				}
-			case <-stopChan:
-				logger.Info("Stopping ping check")
-				return
-			}
-		}
-	}()
-}
-
-func pingWithRetry(tnet *netstack.Net, dst string) error {
-	const (
-		maxAttempts = 5
-		retryDelay  = 2 * time.Second
-	)
-
-	var lastErr error
-	for attempt := 1; attempt <= maxAttempts; attempt++ {
-		logger.Info("Ping attempt %d of %d", attempt, maxAttempts)
-
-		if err := ping(tnet, dst); err != nil {
-			lastErr = err
-			logger.Warn("Ping attempt %d failed: %v", attempt, err)
-
-			if attempt < maxAttempts {
-				time.Sleep(retryDelay)
-				continue
-			}
-			return fmt.Errorf("all ping attempts failed after %d tries, last error: %w",
-				maxAttempts, lastErr)
-		}
-
-		// Successful ping
-		return nil
-	}
-
-	// This shouldn't be reached due to the return in the loop, but added for completeness
-	return fmt.Errorf("unexpected error: all ping attempts failed")
-}
-
-func parseLogLevel(level string) logger.LogLevel {
-	switch strings.ToUpper(level) {
-	case "DEBUG":
-		return logger.DEBUG
-	case "INFO":
-		return logger.INFO
-	case "WARN":
-		return logger.WARN
-	case "ERROR":
-		return logger.ERROR
-	case "FATAL":
-		return logger.FATAL
-	default:
-		return logger.INFO // default to INFO if invalid level provided
-	}
-}
-
-func mapToWireGuardLogLevel(level logger.LogLevel) int {
-	switch level {
-	case logger.DEBUG:
-		return device.LogLevelVerbose
-	// case logger.INFO:
-	// return device.LogLevel
-	case logger.WARN:
-		return device.LogLevelError
-	case logger.ERROR, logger.FATAL:
-		return device.LogLevelSilent
-	default:
-		return device.LogLevelSilent
-	}
-}
-
-func resolveDomain(domain string) (string, error) {
-	// Check if there's a port in the domain
-	host, port, err := net.SplitHostPort(domain)
-	if err != nil {
-		// No port found, use the domain as is
-		host = domain
-		port = ""
-	}
-
-	// Remove any protocol prefix if present
-	if strings.HasPrefix(host, "http://") {
-		host = strings.TrimPrefix(host, "http://")
-	} else if strings.HasPrefix(host, "https://") {
-		host = strings.TrimPrefix(host, "https://")
-	}
-
-	// Lookup IP addresses
-	ips, err := net.LookupIP(host)
-	if err != nil {
-		return "", fmt.Errorf("DNS lookup failed: %v", err)
-	}
-
-	if len(ips) == 0 {
-		return "", fmt.Errorf("no IP addresses found for domain %s", host)
-	}
-
-	// Get the first IPv4 address if available
-	var ipAddr string
-	for _, ip := range ips {
-		if ipv4 := ip.To4(); ipv4 != nil {
-			ipAddr = ipv4.String()
-			break
-		}
-	}
-
-	// If no IPv4 found, use the first IP (might be IPv6)
-	if ipAddr == "" {
-		ipAddr = ips[0].String()
-	}
-
-	// Add port back if it existed
-	if port != "" {
-		ipAddr = net.JoinHostPort(ipAddr, port)
-	}
-
-	return ipAddr, nil
-}
-
 func main() {
-	var (
-		endpoint   string
-		id         string
-		secret     string
-		dns        string
-		privateKey wgtypes.Key
-		err        error
-		logLevel   string
-	)
-
-	// if PANGOLIN_ENDPOINT, NEWT_ID, and NEWT_SECRET are set as environment variables, they will be used as default values
-	endpoint = os.Getenv("PANGOLIN_ENDPOINT")
-	id = os.Getenv("NEWT_ID")
-	secret = os.Getenv("NEWT_SECRET")
-	dns = os.Getenv("DNS")
-	logLevel = os.Getenv("LOG_LEVEL")
-
-	if endpoint == "" {
-		flag.StringVar(&endpoint, "endpoint", "", "Endpoint of your pangolin server")
-	}
-	if id == "" {
-		flag.StringVar(&id, "id", "", "Newt ID")
-	}
-	if secret == "" {
-		flag.StringVar(&secret, "secret", "", "Newt secret")
-	}
-	if dns == "" {
-		flag.StringVar(&dns, "dns", "8.8.8.8", "DNS server to use")
-	}
-	if logLevel == "" {
-		flag.StringVar(&logLevel, "log-level", "INFO", "Log level (DEBUG, INFO, WARN, ERROR, FATAL)")
-	}
-	flag.Parse()
-
-	logger.Init()
-	loggerLevel := parseLogLevel(logLevel)
-	logger.GetLogger().SetLevel(parseLogLevel(logLevel))
-
-	// Validate required fields
-	if endpoint == "" || id == "" || secret == "" {
-		logger.Fatal("endpoint, id, and secret are required either via CLI flags or environment variables")
+	// Check if we're running as a Windows service
+	if isWindowsService() {
+		runService("OlmWireguardService", false, os.Args[1:])
+		fmt.Println("Running as Windows service")
+		return
 	}
 
-	privateKey, err = wgtypes.GeneratePrivateKey()
-	if err != nil {
-		logger.Fatal("Failed to generate private key: %v", err)
-	}
-
-	// Create a new client
-	client, err := websocket.NewClient(
-		id,     // CLI arg takes precedence
-		secret, // CLI arg takes precedence
-		endpoint,
-	)
-	if err != nil {
-		logger.Fatal("Failed to create client: %v", err)
-	}
-
-	// Create TUN device and network stack
-	var tun tun.Device
-	var tnet *netstack.Net
-	var dev *device.Device
-	var pm *proxy.ProxyManager
-	var connected bool
-	var wgData WgData
-
-	client.RegisterHandler("newt/terminate", func(msg websocket.WSMessage) {
-		logger.Info("Received terminate message")
-		if pm != nil {
-			pm.Stop()
+	// Handle service management commands on Windows
+	if runtime.GOOS == "windows" {
+		var command string
+		if len(os.Args) > 1 {
+			command = os.Args[1]
+		} else {
+			command = "default"
 		}
-		if dev != nil {
-			dev.Close()
-		}
-		client.Close()
-	})
 
-	pingStopChan := make(chan struct{})
-	defer close(pingStopChan)
-
-	// Register handlers for different message types
-	client.RegisterHandler("newt/wg/connect", func(msg websocket.WSMessage) {
-		logger.Info("Received registration message")
-
-		if connected {
-			logger.Info("Already connected! But I will send a ping anyway...")
-			// ping(tnet, wgData.ServerIP)
-			err = pingWithRetry(tnet, wgData.ServerIP)
+		switch command {
+		case "install":
+			err := installService()
 			if err != nil {
-				// Handle complete failure after all retries
-				logger.Error("Failed to ping %s: %v", wgData.ServerIP, err)
+				fmt.Printf("Failed to install service: %v\n", err)
+				os.Exit(1)
 			}
+			fmt.Println("Service installed successfully")
 			return
-		}
-
-		jsonData, err := json.Marshal(msg.Data)
-		if err != nil {
-			logger.Info("Error marshaling data: %v", err)
-			return
-		}
-
-		if err := json.Unmarshal(jsonData, &wgData); err != nil {
-			logger.Info("Error unmarshaling target data: %v", err)
-			return
-		}
-
-		logger.Info("Received: %+v", msg)
-		tun, tnet, err = netstack.CreateNetTUN(
-			[]netip.Addr{netip.MustParseAddr(wgData.TunnelIP)},
-			[]netip.Addr{netip.MustParseAddr(dns)},
-			1420)
-		if err != nil {
-			logger.Error("Failed to create TUN device: %v", err)
-		}
-
-		// Create WireGuard device
-		dev = device.NewDevice(tun, conn.NewDefaultBind(), device.NewLogger(
-			mapToWireGuardLogLevel(loggerLevel),
-			"wireguard: ",
-		))
-
-		endpoint, err := resolveDomain(wgData.Endpoint)
-		if err != nil {
-			logger.Error("Failed to resolve endpoint: %v", err)
-			return
-		}
-
-		// Configure WireGuard
-		config := fmt.Sprintf(`private_key=%s
-public_key=%s
-allowed_ip=%s/32
-endpoint=%s
-persistent_keepalive_interval=5`, fixKey(fmt.Sprintf("%s", privateKey)), fixKey(wgData.PublicKey), wgData.ServerIP, endpoint)
-
-		err = dev.IpcSet(config)
-		if err != nil {
-			logger.Error("Failed to configure WireGuard device: %v", err)
-		}
-
-		// Bring up the device
-		err = dev.Up()
-		if err != nil {
-			logger.Error("Failed to bring up WireGuard device: %v", err)
-		}
-
-		logger.Info("WireGuard device created. Lets ping the server now...")
-		// Ping to bring the tunnel up on the server side quickly
-		// ping(tnet, wgData.ServerIP)
-		err = pingWithRetry(tnet, wgData.ServerIP)
-		if err != nil {
-			// Handle complete failure after all retries
-			logger.Error("Failed to ping %s: %v", wgData.ServerIP, err)
-		}
-
-		if !connected {
-			logger.Info("Starting ping check")
-			startPingCheck(tnet, wgData.ServerIP, pingStopChan)
-		}
-
-		// Create proxy manager
-		pm = proxy.NewProxyManager(tnet)
-
-		connected = true
-
-		// add the targets if there are any
-		if len(wgData.Targets.TCP) > 0 {
-			updateTargets(pm, "add", wgData.TunnelIP, "tcp", TargetData{Targets: wgData.Targets.TCP})
-		}
-
-		if len(wgData.Targets.UDP) > 0 {
-			updateTargets(pm, "add", wgData.TunnelIP, "udp", TargetData{Targets: wgData.Targets.UDP})
-		}
-
-		err = pm.Start()
-		if err != nil {
-			logger.Error("Failed to start proxy manager: %v", err)
-		}
-	})
-
-	client.RegisterHandler("newt/tcp/add", func(msg websocket.WSMessage) {
-		logger.Info("Received: %+v", msg)
-
-		// if there is no wgData or pm, we can't add targets
-		if wgData.TunnelIP == "" || pm == nil {
-			logger.Info("No tunnel IP or proxy manager available")
-			return
-		}
-
-		targetData, err := parseTargetData(msg.Data)
-		if err != nil {
-			logger.Info("Error parsing target data: %v", err)
-			return
-		}
-
-		if len(targetData.Targets) > 0 {
-			updateTargets(pm, "add", wgData.TunnelIP, "tcp", targetData)
-		}
-
-		err = pm.Start()
-		if err != nil {
-			logger.Error("Failed to start proxy manager: %v", err)
-		}
-	})
-
-	client.RegisterHandler("newt/udp/add", func(msg websocket.WSMessage) {
-		logger.Info("Received: %+v", msg)
-
-		// if there is no wgData or pm, we can't add targets
-		if wgData.TunnelIP == "" || pm == nil {
-			logger.Info("No tunnel IP or proxy manager available")
-			return
-		}
-
-		targetData, err := parseTargetData(msg.Data)
-		if err != nil {
-			logger.Info("Error parsing target data: %v", err)
-			return
-		}
-
-		if len(targetData.Targets) > 0 {
-			updateTargets(pm, "add", wgData.TunnelIP, "udp", targetData)
-		}
-
-		err = pm.Start()
-		if err != nil {
-			logger.Error("Failed to start proxy manager: %v", err)
-		}
-	})
-
-	client.RegisterHandler("newt/udp/remove", func(msg websocket.WSMessage) {
-		logger.Info("Received: %+v", msg)
-
-		// if there is no wgData or pm, we can't add targets
-		if wgData.TunnelIP == "" || pm == nil {
-			logger.Info("No tunnel IP or proxy manager available")
-			return
-		}
-
-		targetData, err := parseTargetData(msg.Data)
-		if err != nil {
-			logger.Info("Error parsing target data: %v", err)
-			return
-		}
-
-		if len(targetData.Targets) > 0 {
-			updateTargets(pm, "remove", wgData.TunnelIP, "udp", targetData)
-		}
-	})
-
-	client.RegisterHandler("newt/tcp/remove", func(msg websocket.WSMessage) {
-		logger.Info("Received: %+v", msg)
-
-		// if there is no wgData or pm, we can't add targets
-		if wgData.TunnelIP == "" || pm == nil {
-			logger.Info("No tunnel IP or proxy manager available")
-			return
-		}
-
-		targetData, err := parseTargetData(msg.Data)
-		if err != nil {
-			logger.Info("Error parsing target data: %v", err)
-			return
-		}
-
-		if len(targetData.Targets) > 0 {
-			updateTargets(pm, "remove", wgData.TunnelIP, "tcp", targetData)
-		}
-	})
-
-	client.OnConnect(func() error {
-		publicKey := privateKey.PublicKey()
-		logger.Debug("Public key: %s", publicKey)
-
-		err := client.SendMessage("newt/wg/register", map[string]interface{}{
-			"publicKey": fmt.Sprintf("%s", publicKey),
-		})
-		if err != nil {
-			logger.Error("Failed to send registration message: %v", err)
-			return err
-		}
-
-		logger.Info("Sent registration message")
-		return nil
-	})
-
-	// Connect to the WebSocket server
-	if err := client.Connect(); err != nil {
-		logger.Fatal("Failed to connect to server: %v", err)
-	}
-	defer client.Close()
-
-	// Wait for interrupt signal
-	sigCh := make(chan os.Signal, 1)
-	signal.Notify(sigCh, syscall.SIGINT, syscall.SIGTERM)
-	<-sigCh
-
-	// Cleanup
-	dev.Close()
-}
-
-func parseTargetData(data interface{}) (TargetData, error) {
-	var targetData TargetData
-	jsonData, err := json.Marshal(data)
-	if err != nil {
-		logger.Info("Error marshaling data: %v", err)
-		return targetData, err
-	}
-
-	if err := json.Unmarshal(jsonData, &targetData); err != nil {
-		logger.Info("Error unmarshaling target data: %v", err)
-		return targetData, err
-	}
-	return targetData, nil
-}
-
-func updateTargets(pm *proxy.ProxyManager, action string, tunnelIP string, proto string, targetData TargetData) error {
-	for _, t := range targetData.Targets {
-		// Split the first number off of the target with : separator and use as the port
-		parts := strings.Split(t, ":")
-		if len(parts) != 3 {
-			logger.Info("Invalid target format: %s", t)
-			continue
-		}
-
-		// Get the port as an int
-		port := 0
-		_, err := fmt.Sscanf(parts[0], "%d", &port)
-		if err != nil {
-			logger.Info("Invalid port: %s", parts[0])
-			continue
-		}
-
-		if action == "add" {
-			target := parts[1] + ":" + parts[2]
-			// Only remove the specific target if it exists
-			err := pm.RemoveTarget(proto, tunnelIP, port)
+		case "remove", "uninstall":
+			err := removeService()
 			if err != nil {
-				// Ignore "target not found" errors as this is expected for new targets
-				if !strings.Contains(err.Error(), "target not found") {
-					logger.Error("Failed to remove existing target: %v", err)
+				fmt.Printf("Failed to remove service: %v\n", err)
+				os.Exit(1)
+			}
+			fmt.Println("Service removed successfully")
+			return
+		case "start":
+			// Pass the remaining arguments (after "start") to the service
+			serviceArgs := os.Args[2:]
+			err := startService(serviceArgs)
+			if err != nil {
+				fmt.Printf("Failed to start service: %v\n", err)
+				os.Exit(1)
+			}
+			fmt.Println("Service started successfully")
+			return
+		case "stop":
+			err := stopService()
+			if err != nil {
+				fmt.Printf("Failed to stop service: %v\n", err)
+				os.Exit(1)
+			}
+			fmt.Println("Service stopped successfully")
+			return
+		case "status":
+			status, err := getServiceStatus()
+			if err != nil {
+				fmt.Printf("Failed to get service status: %v\n", err)
+				os.Exit(1)
+			}
+			fmt.Printf("Service status: %s\n", status)
+			return
+		case "debug":
+			// get the status and if it is Not Installed then install it first
+			status, err := getServiceStatus()
+			if err != nil {
+				fmt.Printf("Failed to get service status: %v\n", err)
+				os.Exit(1)
+			}
+			if status == "Not Installed" {
+				err := installService()
+				if err != nil {
+					fmt.Printf("Failed to install service: %v\n", err)
+					os.Exit(1)
 				}
+				fmt.Println("Service installed successfully, now running in debug mode")
 			}
 
-			// Add the new target
-			pm.AddTarget(proto, tunnelIP, port, target)
-
-		} else if action == "remove" {
-			logger.Info("Removing target with port %d", port)
-			err := pm.RemoveTarget(proto, tunnelIP, port)
+			// Pass the remaining arguments (after "debug") to the service
+			serviceArgs := os.Args[2:]
+			err = debugService(serviceArgs)
 			if err != nil {
-				logger.Error("Failed to remove target: %v", err)
-				return err
+				fmt.Printf("Failed to debug service: %v\n", err)
+				os.Exit(1)
 			}
+			return
+		case "logs":
+			err := watchLogFile(false)
+			if err != nil {
+				fmt.Printf("Failed to watch log file: %v\n", err)
+				os.Exit(1)
+			}
+			return
+		case "config":
+			if runtime.GOOS == "windows" {
+				showServiceConfig()
+			} else {
+				fmt.Println("Service configuration is only available on Windows")
+			}
+			return
+		case "help", "--help", "-h":
+			fmt.Println("Olm WireGuard VPN Client")
+			fmt.Println("\nWindows Service Management:")
+			fmt.Println("  install     Install the service")
+			fmt.Println("  remove      Remove the service")
+			fmt.Println("  start [args]   Start the service with optional arguments")
+			fmt.Println("  stop        Stop the service")
+			fmt.Println("  status      Show service status")
+			fmt.Println("  debug [args]   Run service in debug mode with optional arguments")
+			fmt.Println("  logs        Tail the service log file")
+			fmt.Println("  config      Show current service configuration")
+			fmt.Println("\nExamples:")
+			fmt.Println("  olm start --enable-http --http-addr :9452")
+			fmt.Println("  olm debug --endpoint https://example.com --id myid --secret mysecret")
+			fmt.Println("\nFor console mode, run without arguments or with standard flags.")
+			return
+		default:
+			// get the status and if it is Not Installed then install it first
+			status, err := getServiceStatus()
+			if err != nil {
+				fmt.Printf("Failed to get service status: %v\n", err)
+				os.Exit(1)
+			}
+			if status == "Not Installed" {
+				err := installService()
+				if err != nil {
+					fmt.Printf("Failed to install service: %v\n", err)
+					os.Exit(1)
+				}
+				fmt.Println("Service installed successfully, now running")
+			}
+
+			// Pass the remaining arguments (after "debug") to the service
+			serviceArgs := os.Args[1:]
+			err = debugService(serviceArgs)
+			if err != nil {
+				fmt.Printf("Failed to debug service: %v\n", err)
+				os.Exit(1)
+			}
+			return
 		}
 	}
 
-	return nil
+	// Create a context that will be cancelled on interrupt signals
+	signalCtx, stop := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
+	defer stop()
+
+	// Create a separate context for programmatic shutdown (e.g., via API exit)
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// Run in console mode
+	runOlmMainWithArgs(ctx, cancel, signalCtx, os.Args[1:])
+}
+
+func runOlmMainWithArgs(ctx context.Context, cancel context.CancelFunc, signalCtx context.Context, args []string) {
+	// Setup Windows event logging if on Windows
+	if runtime.GOOS == "windows" {
+		setupWindowsEventLog()
+	} else {
+		// Initialize logger for non-Windows platforms
+		logger.Init(nil)
+	}
+
+	// Load configuration from file, env vars, and CLI args
+	// Priority: CLI args > Env vars > Config file > Defaults
+	// Use the passed args parameter instead of os.Args[1:] to support Windows service mode
+	config, showVersion, showConfig, err := LoadConfig(args)
+	if err != nil {
+		fmt.Printf("Failed to load configuration: %v\n", err)
+		return
+	}
+
+	// Handle --show-config flag
+	if showConfig {
+		config.ShowConfig()
+		os.Exit(0)
+	}
+
+	olmVersion := "version_replaceme"
+	if showVersion {
+		fmt.Println("Olm version " + olmVersion)
+		os.Exit(0)
+	}
+	logger.Info("Olm version %s", olmVersion)
+
+	config.Version = olmVersion
+
+	if err := SaveConfig(config); err != nil {
+		logger.Error("Failed to save full olm config: %v", err)
+	} else {
+		logger.Debug("Saved full olm config with all options")
+	}
+
+	if err := updates.CheckForUpdate("fosrl", "olm", config.Version); err != nil {
+		logger.Debug("Failed to check for updates: %v", err)
+	}
+
+	// Create a new olm.Config struct and copy values from the main config
+	olmConfig := olm.GlobalConfig{
+		LogLevel:     config.LogLevel,
+		EnableAPI:    config.EnableAPI,
+		HTTPAddr:     config.HTTPAddr,
+		SocketPath:   config.SocketPath,
+		Version:      config.Version,
+		Agent:        "Olm CLI",
+		OnExit:       cancel, // Pass cancel function directly to trigger shutdown
+		OnTerminated: cancel,
+	}
+
+	olm.Init(ctx, olmConfig)
+	if err := olm.StartApi(); err != nil {
+		logger.Fatal("Failed to start API server: %v", err)
+	}
+
+	if config.ID != "" && config.Secret != "" && config.Endpoint != "" {
+		tunnelConfig := olm.TunnelConfig{
+			Endpoint:             config.Endpoint,
+			ID:                   config.ID,
+			Secret:               config.Secret,
+			UserToken:            config.UserToken,
+			MTU:                  config.MTU,
+			DNS:                  config.DNS,
+			UpstreamDNS:          config.UpstreamDNS,
+			InterfaceName:        config.InterfaceName,
+			Holepunch:            !config.DisableHolepunch,
+			TlsClientCert:        config.TlsClientCert,
+			PingIntervalDuration: config.PingIntervalDuration,
+			PingTimeoutDuration:  config.PingTimeoutDuration,
+			OrgID:                config.OrgID,
+			OverrideDNS:          config.OverrideDNS,
+			DisableRelay:         config.DisableRelay,
+			EnableUAPI:           true,
+		}
+		go olm.StartTunnel(tunnelConfig)
+	} else {
+		logger.Info("Incomplete tunnel configuration, not starting tunnel")
+	}
+
+	// Wait for either signal or programmatic shutdown
+	select {
+	case <-signalCtx.Done():
+		logger.Info("Shutdown signal received, cleaning up...")
+	case <-ctx.Done():
+		logger.Info("Shutdown requested via API, cleaning up...")
+	}
+
+	// Clean up resources
+	olm.Close()
+	logger.Info("Shutdown complete")
 }

namespace.sh

@@ -0,0 +1,126 @@
+#!/bin/bash
+
+# Configuration
+NS_NAME="isolated_ns"       # Name of the namespace
+VETH_HOST="veth_host"       # Interface name on host side
+VETH_NS="veth_ns"           # Interface name inside namespace
+HOST_IP="192.168.15.1"      # Gateway IP for the namespace (host side)
+NS_IP="192.168.15.2"        # IP address for the namespace
+SUBNET_CIDR="24"            # Subnet mask
+DNS_SERVER="8.8.8.8"        # DNS to use inside namespace
+
+# Detect the main physical interface (gateway to internet)
+PHY_IFACE=$(ip route get 8.8.8.8 | awk -- '{printf $5}')
+
+# Helper function to check for root
+check_root() {
+    if [ "$EUID" -ne 0 ]; then
+        echo "Error: This script must be run as root."
+        exit 1
+    fi
+}
+
+setup_ns() {
+    echo "Bringing up namespace '$NS_NAME'..."
+
+    # 1. Create the network namespace
+    if ip netns list | grep -q "$NS_NAME"; then
+        echo "Namespace $NS_NAME already exists. Run 'down' first."
+        exit 1
+    fi
+    ip netns add "$NS_NAME"
+
+    # 2. Create veth pair
+    ip link add "$VETH_HOST" type veth peer name "$VETH_NS"
+
+    # 3. Move peer interface to namespace
+    ip link set "$VETH_NS" netns "$NS_NAME"
+
+    # 4. Configure Host Side Interface
+    ip addr add "${HOST_IP}/${SUBNET_CIDR}" dev "$VETH_HOST"
+    ip link set "$VETH_HOST" up
+
+    # 5. Configure Namespace Side Interface
+    ip netns exec "$NS_NAME" ip addr add "${NS_IP}/${SUBNET_CIDR}" dev "$VETH_NS"
+    ip netns exec "$NS_NAME" ip link set "$VETH_NS" up
+    
+    # 6. Bring up loopback inside namespace (crucial for many apps)
+    ip netns exec "$NS_NAME" ip link set lo up
+
+    # 7. Routing: Add default gateway inside namespace pointing to host
+    ip netns exec "$NS_NAME" ip route add default via "$HOST_IP"
+
+    # 8. Enable IP forwarding on host
+    echo 1 > /proc/sys/net/ipv4/ip_forward
+
+    # 9. NAT/Masquerade: Allow traffic from namespace to go out physical interface
+    # We verify rule doesn't exist first to avoid duplicates
+    iptables -t nat -C POSTROUTING -s "${NS_IP}/${SUBNET_CIDR}" -o "$PHY_IFACE" -j MASQUERADE 2>/dev/null || \
+    iptables -t nat -A POSTROUTING -s "${NS_IP}/${SUBNET_CIDR}" -o "$PHY_IFACE" -j MASQUERADE
+
+    # Allow forwarding from host veth to WAN and back
+    iptables -C FORWARD -i "$VETH_HOST" -o "$PHY_IFACE" -j ACCEPT 2>/dev/null || \
+    iptables -A FORWARD -i "$VETH_HOST" -o "$PHY_IFACE" -j ACCEPT
+
+    iptables -C FORWARD -i "$PHY_IFACE" -o "$VETH_HOST" -j ACCEPT 2>/dev/null || \
+    iptables -A FORWARD -i "$PHY_IFACE" -o "$VETH_HOST" -j ACCEPT
+
+    # 10. DNS Setup
+    # Netns uses /etc/netns/<name>/resolv.conf if it exists
+    mkdir -p "/etc/netns/$NS_NAME"
+    echo "nameserver $DNS_SERVER" > "/etc/netns/$NS_NAME/resolv.conf"
+
+    echo "Namespace $NS_NAME is UP."
+    echo "To enter shell: sudo ip netns exec $NS_NAME bash"
+}
+
+teardown_ns() {
+    echo "Tearing down namespace '$NS_NAME'..."
+
+    # 1. Remove Namespace (this automatically deletes the veth pair inside it)
+    # The host side veth usually disappears when the peer is destroyed.
+    if ip netns list | grep -q "$NS_NAME"; then
+        ip netns del "$NS_NAME"
+    else
+        echo "Namespace $NS_NAME does not exist."
+    fi
+
+    # 2. Clean up veth host side if it still lingers
+    if ip link show "$VETH_HOST" > /dev/null 2>&1; then
+        ip link delete "$VETH_HOST"
+    fi
+
+    # 3. Remove iptables rules
+    # We use -D to delete the specific rules we added
+    iptables -t nat -D POSTROUTING -s "${NS_IP}/${SUBNET_CIDR}" -o "$PHY_IFACE" -j MASQUERADE 2>/dev/null
+    iptables -D FORWARD -i "$VETH_HOST" -o "$PHY_IFACE" -j ACCEPT 2>/dev/null
+    iptables -D FORWARD -i "$PHY_IFACE" -o "$VETH_HOST" -j ACCEPT 2>/dev/null
+
+    # 4. Remove DNS config
+    rm -rf "/etc/netns/$NS_NAME"
+
+    echo "Namespace $NS_NAME is DOWN."
+}
+
+test_connectivity() {
+    echo "Testing connectivity inside $NS_NAME..."
+    ip netns exec "$NS_NAME" ping -c 3 8.8.8.8
+}
+
+# Main execution logic
+check_root
+
+case "$1" in
+    up)
+        setup_ns
+        ;;
+    down)
+        teardown_ns
+        ;;
+    test)
+        test_connectivity
+        ;;
+    *)
+        echo "Usage: $0 {up|down|test}"
+        exit 1
+esac

olm.iss

@@ -0,0 +1,152 @@
+; Script generated by the Inno Setup Script Wizard.
+; SEE THE DOCUMENTATION FOR DETAILS ON CREATING INNO SETUP SCRIPT FILES!
+
+#define MyAppName "olm"
+#define MyAppVersion "1.0.0"
+#define MyAppPublisher "Fossorial Inc."
+#define MyAppURL "https://pangolin.net"
+#define MyAppExeName "olm.exe"
+
+[Setup]
+; NOTE: The value of AppId uniquely identifies this application. Do not use the same AppId value in installers for other applications.
+; (To generate a new GUID, click Tools | Generate GUID inside the IDE.)
+AppId={{44A24E4C-B616-476F-ADE7-8D56B930959E}
+AppName={#MyAppName}
+AppVersion={#MyAppVersion}
+;AppVerName={#MyAppName} {#MyAppVersion}
+AppPublisher={#MyAppPublisher}
+AppPublisherURL={#MyAppURL}
+AppSupportURL={#MyAppURL}
+AppUpdatesURL={#MyAppURL}
+DefaultDirName={autopf}\{#MyAppName}
+UninstallDisplayIcon={app}\{#MyAppExeName}
+; "ArchitecturesAllowed=x64compatible" specifies that Setup cannot run
+; on anything but x64 and Windows 11 on Arm.
+ArchitecturesAllowed=x64compatible
+; "ArchitecturesInstallIn64BitMode=x64compatible" requests that the
+; install be done in "64-bit mode" on x64 or Windows 11 on Arm,
+; meaning it should use the native 64-bit Program Files directory and
+; the 64-bit view of the registry.
+ArchitecturesInstallIn64BitMode=x64compatible
+DefaultGroupName={#MyAppName}
+DisableProgramGroupPage=yes
+; Uncomment the following line to run in non administrative install mode (install for current user only).
+;PrivilegesRequired=lowest
+OutputBaseFilename=mysetup
+SolidCompression=yes
+WizardStyle=modern
+; Add this to ensure PATH changes are applied and the system is prompted for a restart if needed
+RestartIfNeededByRun=no
+ChangesEnvironment=true
+
+[Languages]
+Name: "english"; MessagesFile: "compiler:Default.isl"
+
+[Files]
+; The 'DestName' flag ensures that 'olm_windows_amd64.exe' is installed as 'olm.exe'
+Source: "Z:\olm_windows_amd64.exe"; DestDir: "{app}"; DestName: "{#MyAppExeName}"; Flags: ignoreversion
+Source: "Z:\wintun.dll"; DestDir: "{app}"; Flags: ignoreversion
+; NOTE: Don't use "Flags: ignoreversion" on any shared system files
+
+[Icons]
+Name: "{group}\{#MyAppName}"; Filename: "{app}\{#MyAppExeName}"
+
+[Registry]
+; Add the application's installation directory to the system PATH environment variable.
+; HKLM (HKEY_LOCAL_MACHINE) is used for system-wide changes.
+; The 'Path' variable is located under 'SYSTEM\CurrentControlSet\Control\Session Manager\Environment'.
+; ValueType: expandsz allows for environment variables (like %ProgramFiles%) in the path.
+; ValueData: "{olddata};{app}" appends the current application directory to the existing PATH.
+; Note: Removal during uninstallation is handled by CurUninstallStepChanged procedure in [Code] section.
+; Check: NeedsAddPath ensures this is applied only if the path is not already present.
+[Registry]
+; Add the application's installation directory to the system PATH.
+Root: HKLM; Subkey: "SYSTEM\CurrentControlSet\Control\Session Manager\Environment"; \
+    ValueType: expandsz; ValueName: "Path"; ValueData: "{olddata};{app}"; \
+    Check: NeedsAddPath(ExpandConstant('{app}'))
+
+[Code]
+function NeedsAddPath(Path: string): boolean;
+var
+  OrigPath: string;
+begin
+  if not RegQueryStringValue(HKEY_LOCAL_MACHINE,
+    'SYSTEM\CurrentControlSet\Control\Session Manager\Environment',
+    'Path', OrigPath)
+  then begin
+    // Path variable doesn't exist at all, so we definitely need to add it.
+    Result := True;
+    exit;
+  end;
+
+  // Perform a case-insensitive check to see if the path is already present.
+  // We add semicolons to prevent partial matches (e.g., matching C:\App in C:\App2).
+  if Pos(';' + UpperCase(Path) + ';', ';' + UpperCase(OrigPath) + ';') > 0 then
+    Result := False
+  else
+    Result := True;
+end;
+
+procedure RemovePathEntry(PathToRemove: string);
+var
+  OrigPath: string;
+  NewPath: string;
+  PathList: TStringList;
+  I: Integer;
+begin
+  if not RegQueryStringValue(HKEY_LOCAL_MACHINE,
+    'SYSTEM\CurrentControlSet\Control\Session Manager\Environment',
+    'Path', OrigPath)
+  then begin
+    // Path variable doesn't exist, nothing to remove
+    exit;
+  end;
+
+  // Create a string list to parse the PATH entries
+  PathList := TStringList.Create;
+  try
+    // Split the PATH by semicolons
+    PathList.Delimiter := ';';
+    PathList.StrictDelimiter := True;
+    PathList.DelimitedText := OrigPath;
+
+    // Find and remove the matching entry (case-insensitive)
+    for I := PathList.Count - 1 downto 0 do
+    begin
+      if CompareText(Trim(PathList[I]), Trim(PathToRemove)) = 0 then
+      begin
+        Log('Found and removing PATH entry: ' + PathList[I]);
+        PathList.Delete(I);
+      end;
+    end;
+
+    // Reconstruct the PATH
+    NewPath := PathList.DelimitedText;
+
+    // Write the new PATH back to the registry
+    if RegWriteExpandStringValue(HKEY_LOCAL_MACHINE,
+      'SYSTEM\CurrentControlSet\Control\Session Manager\Environment',
+      'Path', NewPath)
+    then
+      Log('Successfully removed path entry: ' + PathToRemove)
+    else
+      Log('Failed to write modified PATH to registry');
+  finally
+    PathList.Free;
+  end;
+end;
+
+procedure CurUninstallStepChanged(CurUninstallStep: TUninstallStep);
+var
+  AppPath: string;
+begin
+  if CurUninstallStep = usUninstall then
+  begin
+    // Get the application installation path
+    AppPath := ExpandConstant('{app}');
+    Log('Removing PATH entry for: ' + AppPath);
+
+    // Remove only our path entry from the system PATH
+    RemovePathEntry(AppPath);
+  end;
+end;

olm/types.go

@@ -0,0 +1,67 @@
+package olm
+
+import (
+	"time"
+
+	"github.com/fosrl/olm/peers"
+)
+
+type WgData struct {
+	Sites         []peers.SiteConfig `json:"sites"`
+	TunnelIP      string             `json:"tunnelIP"`
+	UtilitySubnet string             `json:"utilitySubnet"` // this is for things like the DNS server, and alias addresses
+}
+
+type GlobalConfig struct {
+	// Logging
+	LogLevel string
+
+	// HTTP server
+	EnableAPI  bool
+	HTTPAddr   string
+	SocketPath string
+	Version    string
+	Agent      string
+
+	// Callbacks
+	OnRegistered func()
+	OnConnected  func()
+	OnTerminated func()
+	OnAuthError  func(statusCode int, message string) // Called when auth fails (401/403)
+	OnExit       func()                               // Called when exit is requested via API
+}
+
+type TunnelConfig struct {
+	// Connection settings
+	Endpoint  string
+	ID        string
+	Secret    string
+	UserToken string
+
+	// Network settings
+	MTU           int
+	DNS           string
+	UpstreamDNS   []string
+	InterfaceName string
+
+	// Advanced
+	Holepunch     bool
+	TlsClientCert string
+
+	// Parsed values (not in JSON)
+	PingIntervalDuration time.Duration
+	PingTimeoutDuration  time.Duration
+
+	OrgID string
+	// DoNotCreateNewClient bool
+
+	FileDescriptorTun  uint32
+	FileDescriptorUAPI uint32
+
+	EnableUAPI bool
+
+	OverrideDNS bool
+	TunnelDNS   bool
+
+	DisableRelay bool
+}

olm/util.go

@@ -0,0 +1,55 @@
+package olm
+
+import (
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/network"
+	"github.com/fosrl/olm/websocket"
+)
+
+func sendPing(olm *websocket.Client) error {
+	err := olm.SendMessage("olm/ping", map[string]interface{}{
+		"timestamp": time.Now().Unix(),
+		"userToken": olm.GetConfig().UserToken,
+	})
+	if err != nil {
+		logger.Error("Failed to send ping message: %v", err)
+		return err
+	}
+	logger.Debug("Sent ping message")
+	return nil
+}
+
+func keepSendingPing(olm *websocket.Client) {
+	// Send ping immediately on startup
+	if err := sendPing(olm); err != nil {
+		logger.Error("Failed to send initial ping: %v", err)
+	} else {
+		logger.Info("Sent initial ping message")
+	}
+
+	// Set up ticker for one minute intervals
+	ticker := time.NewTicker(1 * time.Minute)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-stopPing:
+			logger.Info("Stopping ping messages")
+			return
+		case <-ticker.C:
+			if err := sendPing(olm); err != nil {
+				logger.Error("Failed to send periodic ping: %v", err)
+			}
+		}
+	}
+}
+
+func GetNetworkSettingsJSON() (string, error) {
+	return network.GetJSON()
+}
+
+func GetNetworkSettingsIncrementor() int {
+	return network.GetIncrementor()
+}

peers/manager.go

@@ -0,0 +1,888 @@
+package peers
+
+import (
+	"fmt"
+	"net"
+	"strconv"
+	"strings"
+	"sync"
+
+	"github.com/fosrl/newt/bind"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/network"
+	"github.com/fosrl/newt/util"
+	"github.com/fosrl/olm/api"
+	olmDevice "github.com/fosrl/olm/device"
+	"github.com/fosrl/olm/dns"
+	"github.com/fosrl/olm/peers/monitor"
+	"github.com/fosrl/olm/websocket"
+	"golang.zx2c4.com/wireguard/device"
+	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+)
+
+// PeerManagerConfig contains the configuration for creating a PeerManager
+type PeerManagerConfig struct {
+	Device        *device.Device
+	DNSProxy      *dns.DNSProxy
+	InterfaceName string
+	PrivateKey    wgtypes.Key
+	// For peer monitoring
+	MiddleDev  *olmDevice.MiddleDevice
+	LocalIP    string
+	SharedBind *bind.SharedBind
+	// WSClient is optional - if nil, relay messages won't be sent
+	WSClient  *websocket.Client
+	APIServer *api.API
+}
+
+type PeerManager struct {
+	mu            sync.RWMutex
+	device        *device.Device
+	peers         map[int]SiteConfig
+	peerMonitor   *monitor.PeerMonitor
+	dnsProxy      *dns.DNSProxy
+	interfaceName string
+	privateKey    wgtypes.Key
+	// allowedIPOwners tracks which peer currently "owns" each allowed IP in WireGuard
+	// key is the CIDR string, value is the siteId that has it configured in WG
+	allowedIPOwners map[string]int
+	// allowedIPClaims tracks all peers that claim each allowed IP
+	// key is the CIDR string, value is a set of siteIds that want this IP
+	allowedIPClaims map[string]map[int]bool
+	APIServer       *api.API
+}
+
+// NewPeerManager creates a new PeerManager with an internal PeerMonitor
+func NewPeerManager(config PeerManagerConfig) *PeerManager {
+	pm := &PeerManager{
+		device:          config.Device,
+		peers:           make(map[int]SiteConfig),
+		dnsProxy:        config.DNSProxy,
+		interfaceName:   config.InterfaceName,
+		privateKey:      config.PrivateKey,
+		allowedIPOwners: make(map[string]int),
+		allowedIPClaims: make(map[string]map[int]bool),
+		APIServer:       config.APIServer,
+	}
+
+	// Create the peer monitor
+	pm.peerMonitor = monitor.NewPeerMonitor(
+		config.WSClient,
+		config.MiddleDev,
+		config.LocalIP,
+		config.SharedBind,
+		config.APIServer,
+	)
+
+	return pm
+}
+
+func (pm *PeerManager) GetPeer(siteId int) (SiteConfig, bool) {
+	pm.mu.RLock()
+	defer pm.mu.RUnlock()
+	peer, ok := pm.peers[siteId]
+	return peer, ok
+}
+
+func (pm *PeerManager) GetAllPeers() []SiteConfig {
+	pm.mu.RLock()
+	defer pm.mu.RUnlock()
+	peers := make([]SiteConfig, 0, len(pm.peers))
+	for _, peer := range pm.peers {
+		peers = append(peers, peer)
+	}
+	return peers
+}
+
+func (pm *PeerManager) AddPeer(siteConfig SiteConfig) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	// build the allowed IPs list from the remote subnets and aliases and add them to the peer
+	allowedIPs := make([]string, 0, len(siteConfig.RemoteSubnets)+len(siteConfig.Aliases))
+	allowedIPs = append(allowedIPs, siteConfig.RemoteSubnets...)
+	for _, alias := range siteConfig.Aliases {
+		allowedIPs = append(allowedIPs, alias.AliasAddress+"/32")
+	}
+	siteConfig.AllowedIps = allowedIPs
+
+	// Register claims for all allowed IPs and determine which ones this peer will own
+	ownedIPs := make([]string, 0, len(allowedIPs))
+	for _, ip := range allowedIPs {
+		pm.claimAllowedIP(siteConfig.SiteId, ip)
+		// Check if this peer became the owner
+		if pm.allowedIPOwners[ip] == siteConfig.SiteId {
+			ownedIPs = append(ownedIPs, ip)
+		}
+	}
+
+	// Create a config with only the owned IPs for WireGuard
+	wgConfig := siteConfig
+	wgConfig.AllowedIps = ownedIPs
+
+	if err := ConfigurePeer(pm.device, wgConfig, pm.privateKey, pm.peerMonitor.IsPeerRelayed(siteConfig.SiteId)); err != nil {
+		return err
+	}
+
+	if err := network.AddRouteForServerIP(siteConfig.ServerIP, pm.interfaceName); err != nil {
+		logger.Error("Failed to add route for server IP: %v", err)
+	}
+	if err := network.AddRoutes(siteConfig.RemoteSubnets, pm.interfaceName); err != nil {
+		logger.Error("Failed to add routes for remote subnets: %v", err)
+	}
+	for _, alias := range siteConfig.Aliases {
+		address := net.ParseIP(alias.AliasAddress)
+		if address == nil {
+			continue
+		}
+		pm.dnsProxy.AddDNSRecord(alias.Alias, address)
+	}
+
+	monitorAddress := strings.Split(siteConfig.ServerIP, "/")[0]
+	monitorPeer := net.JoinHostPort(monitorAddress, strconv.Itoa(int(siteConfig.ServerPort+1))) // +1 for the monitor port
+
+	err := pm.peerMonitor.AddPeer(siteConfig.SiteId, monitorPeer, siteConfig.Endpoint) // always use the real site endpoint for hole punch monitoring
+	if err != nil {
+		logger.Warn("Failed to setup monitoring for site %d: %v", siteConfig.SiteId, err)
+	} else {
+		logger.Info("Started monitoring for site %d at %s", siteConfig.SiteId, monitorPeer)
+	}
+
+	pm.peers[siteConfig.SiteId] = siteConfig
+
+	pm.APIServer.AddPeerStatus(siteConfig.SiteId, siteConfig.Name, false, 0, siteConfig.Endpoint, false)
+
+	// Perform rapid initial holepunch test (outside of lock to avoid blocking)
+	// This quickly determines if holepunch is viable and triggers relay if not
+	go pm.performRapidInitialTest(siteConfig.SiteId, siteConfig.Endpoint)
+
+	return nil
+}
+
+func (pm *PeerManager) RemovePeer(siteId int) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	if err := RemovePeer(pm.device, siteId, peer.PublicKey); err != nil {
+		return err
+	}
+
+	if err := network.RemoveRouteForServerIP(peer.ServerIP, pm.interfaceName); err != nil {
+		logger.Error("Failed to remove route for server IP: %v", err)
+	}
+
+	// Only remove routes for subnets that aren't used by other peers
+	for _, subnet := range peer.RemoteSubnets {
+		subnetStillInUse := false
+		for otherSiteId, otherPeer := range pm.peers {
+			if otherSiteId == siteId {
+				continue // Skip the peer being removed
+			}
+			for _, otherSubnet := range otherPeer.RemoteSubnets {
+				if otherSubnet == subnet {
+					subnetStillInUse = true
+					break
+				}
+			}
+			if subnetStillInUse {
+				break
+			}
+		}
+		if !subnetStillInUse {
+			if err := network.RemoveRoutes([]string{subnet}); err != nil {
+				logger.Error("Failed to remove route for remote subnet %s: %v", subnet, err)
+			}
+		}
+	}
+
+	// For aliases
+	for _, alias := range peer.Aliases {
+		address := net.ParseIP(alias.AliasAddress)
+		if address == nil {
+			continue
+		}
+		pm.dnsProxy.RemoveDNSRecord(alias.Alias, address)
+	}
+
+	// Release all IP claims and promote other peers as needed
+	// Collect promotions first to avoid modifying while iterating
+	type promotion struct {
+		newOwner int
+		cidr     string
+	}
+	var promotions []promotion
+
+	for _, ip := range peer.AllowedIps {
+		newOwner, promoted := pm.releaseAllowedIP(siteId, ip)
+		if promoted && newOwner >= 0 {
+			promotions = append(promotions, promotion{newOwner: newOwner, cidr: ip})
+		}
+	}
+
+	// Apply promotions - update WireGuard config for newly promoted peers
+	// Group by peer to avoid multiple config updates
+	promotedPeers := make(map[int]bool)
+	for _, p := range promotions {
+		promotedPeers[p.newOwner] = true
+		logger.Info("Promoted peer %d to owner of IP %s", p.newOwner, p.cidr)
+	}
+
+	for promotedPeerId := range promotedPeers {
+		if promotedPeer, exists := pm.peers[promotedPeerId]; exists {
+			// Build the list of IPs this peer now owns
+			ownedIPs := pm.getOwnedAllowedIPs(promotedPeerId)
+			wgConfig := promotedPeer
+			wgConfig.AllowedIps = ownedIPs
+			if err := ConfigurePeer(pm.device, wgConfig, pm.privateKey, pm.peerMonitor.IsPeerRelayed(promotedPeerId)); err != nil {
+				logger.Error("Failed to update promoted peer %d: %v", promotedPeerId, err)
+			}
+		}
+	}
+
+	// Stop monitoring this peer
+	pm.peerMonitor.RemovePeer(siteId)
+	logger.Info("Stopped monitoring for site %d", siteId)
+
+	pm.APIServer.RemovePeerStatus(siteId)
+
+	delete(pm.peers, siteId)
+	return nil
+}
+
+func (pm *PeerManager) UpdatePeer(siteConfig SiteConfig) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	oldPeer, exists := pm.peers[siteConfig.SiteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteConfig.SiteId)
+	}
+
+	// If public key changed, remove old peer first
+	if siteConfig.PublicKey != oldPeer.PublicKey {
+		if err := RemovePeer(pm.device, siteConfig.SiteId, oldPeer.PublicKey); err != nil {
+			logger.Error("Failed to remove old peer: %v", err)
+		}
+	}
+
+	// Build the new allowed IPs list
+	newAllowedIPs := make([]string, 0, len(siteConfig.RemoteSubnets)+len(siteConfig.Aliases))
+	newAllowedIPs = append(newAllowedIPs, siteConfig.RemoteSubnets...)
+	for _, alias := range siteConfig.Aliases {
+		newAllowedIPs = append(newAllowedIPs, alias.AliasAddress+"/32")
+	}
+	siteConfig.AllowedIps = newAllowedIPs
+
+	// Handle allowed IP claim changes
+	oldAllowedIPs := make(map[string]bool)
+	for _, ip := range oldPeer.AllowedIps {
+		oldAllowedIPs[ip] = true
+	}
+	newAllowedIPsSet := make(map[string]bool)
+	for _, ip := range newAllowedIPs {
+		newAllowedIPsSet[ip] = true
+	}
+
+	// Track peers that need WireGuard config updates due to promotions
+	peersToUpdate := make(map[int]bool)
+
+	// Release claims for removed IPs and handle promotions
+	for ip := range oldAllowedIPs {
+		if !newAllowedIPsSet[ip] {
+			newOwner, promoted := pm.releaseAllowedIP(siteConfig.SiteId, ip)
+			if promoted && newOwner >= 0 {
+				peersToUpdate[newOwner] = true
+				logger.Info("Promoted peer %d to owner of IP %s", newOwner, ip)
+			}
+		}
+	}
+
+	// Add claims for new IPs
+	for ip := range newAllowedIPsSet {
+		if !oldAllowedIPs[ip] {
+			pm.claimAllowedIP(siteConfig.SiteId, ip)
+		}
+	}
+
+	// Build the list of IPs this peer owns for WireGuard config
+	ownedIPs := pm.getOwnedAllowedIPs(siteConfig.SiteId)
+	wgConfig := siteConfig
+	wgConfig.AllowedIps = ownedIPs
+
+	if err := ConfigurePeer(pm.device, wgConfig, pm.privateKey, pm.peerMonitor.IsPeerRelayed(siteConfig.SiteId)); err != nil {
+		return err
+	}
+
+	// Update WireGuard config for any promoted peers
+	for promotedPeerId := range peersToUpdate {
+		if promotedPeer, exists := pm.peers[promotedPeerId]; exists {
+			promotedOwnedIPs := pm.getOwnedAllowedIPs(promotedPeerId)
+			promotedWgConfig := promotedPeer
+			promotedWgConfig.AllowedIps = promotedOwnedIPs
+			if err := ConfigurePeer(pm.device, promotedWgConfig, pm.privateKey, pm.peerMonitor.IsPeerRelayed(promotedPeerId)); err != nil {
+				logger.Error("Failed to update promoted peer %d: %v", promotedPeerId, err)
+			}
+		}
+	}
+
+	// Handle remote subnet route changes
+	// Calculate added and removed subnets
+	oldSubnets := make(map[string]bool)
+	for _, s := range oldPeer.RemoteSubnets {
+		oldSubnets[s] = true
+	}
+	newSubnets := make(map[string]bool)
+	for _, s := range siteConfig.RemoteSubnets {
+		newSubnets[s] = true
+	}
+
+	var addedSubnets []string
+	var removedSubnets []string
+
+	for s := range newSubnets {
+		if !oldSubnets[s] {
+			addedSubnets = append(addedSubnets, s)
+		}
+	}
+	for s := range oldSubnets {
+		if !newSubnets[s] {
+			removedSubnets = append(removedSubnets, s)
+		}
+	}
+
+	// Remove routes for removed subnets (only if no other peer needs them)
+	for _, subnet := range removedSubnets {
+		subnetStillInUse := false
+		for otherSiteId, otherPeer := range pm.peers {
+			if otherSiteId == siteConfig.SiteId {
+				continue // Skip the current peer (already updated)
+			}
+			for _, otherSubnet := range otherPeer.RemoteSubnets {
+				if otherSubnet == subnet {
+					subnetStillInUse = true
+					break
+				}
+			}
+			if subnetStillInUse {
+				break
+			}
+		}
+		if !subnetStillInUse {
+			if err := network.RemoveRoutes([]string{subnet}); err != nil {
+				logger.Error("Failed to remove route for subnet %s: %v", subnet, err)
+			}
+		}
+	}
+
+	// Add routes for added subnets
+	if len(addedSubnets) > 0 {
+		if err := network.AddRoutes(addedSubnets, pm.interfaceName); err != nil {
+			logger.Error("Failed to add routes: %v", err)
+		}
+	}
+
+	// Update aliases
+	// Remove old aliases
+	for _, alias := range oldPeer.Aliases {
+		address := net.ParseIP(alias.AliasAddress)
+		if address == nil {
+			continue
+		}
+		pm.dnsProxy.RemoveDNSRecord(alias.Alias, address)
+	}
+	// Add new aliases
+	for _, alias := range siteConfig.Aliases {
+		address := net.ParseIP(alias.AliasAddress)
+		if address == nil {
+			continue
+		}
+		pm.dnsProxy.AddDNSRecord(alias.Alias, address)
+	}
+
+	pm.peerMonitor.UpdateHolepunchEndpoint(siteConfig.SiteId, siteConfig.Endpoint)
+
+	monitorAddress := strings.Split(siteConfig.ServerIP, "/")[0]
+	monitorPeer := net.JoinHostPort(monitorAddress, strconv.Itoa(int(siteConfig.ServerPort+1))) // +1 for the monitor port
+	pm.peerMonitor.UpdatePeerEndpoint(siteConfig.SiteId, monitorPeer)                           // +1 for monitor port
+
+	pm.peers[siteConfig.SiteId] = siteConfig
+	return nil
+}
+
+// claimAllowedIP registers a peer's claim to an allowed IP.
+// If no other peer owns it in WireGuard, this peer becomes the owner.
+// Must be called with lock held.
+func (pm *PeerManager) claimAllowedIP(siteId int, cidr string) {
+	// Add to claims
+	if pm.allowedIPClaims[cidr] == nil {
+		pm.allowedIPClaims[cidr] = make(map[int]bool)
+	}
+	pm.allowedIPClaims[cidr][siteId] = true
+
+	// If no owner yet, this peer becomes the owner
+	if _, hasOwner := pm.allowedIPOwners[cidr]; !hasOwner {
+		pm.allowedIPOwners[cidr] = siteId
+	}
+}
+
+// releaseAllowedIP removes a peer's claim to an allowed IP.
+// If this peer was the owner, it promotes another claimant to owner.
+// Returns the new owner's siteId (or -1 if no new owner) and whether promotion occurred.
+// Must be called with lock held.
+func (pm *PeerManager) releaseAllowedIP(siteId int, cidr string) (newOwner int, promoted bool) {
+	// Remove from claims
+	if claims, exists := pm.allowedIPClaims[cidr]; exists {
+		delete(claims, siteId)
+		if len(claims) == 0 {
+			delete(pm.allowedIPClaims, cidr)
+		}
+	}
+
+	// Check if this peer was the owner
+	owner, isOwned := pm.allowedIPOwners[cidr]
+	if !isOwned || owner != siteId {
+		return -1, false // Not the owner, nothing to promote
+	}
+
+	// This peer was the owner, need to find a new owner
+	delete(pm.allowedIPOwners, cidr)
+
+	// Find another claimant to promote
+	if claims, exists := pm.allowedIPClaims[cidr]; exists && len(claims) > 0 {
+		for claimantId := range claims {
+			pm.allowedIPOwners[cidr] = claimantId
+			return claimantId, true
+		}
+	}
+
+	return -1, false
+}
+
+// getOwnedAllowedIPs returns the list of allowed IPs that a peer currently owns in WireGuard.
+// Must be called with lock held.
+func (pm *PeerManager) getOwnedAllowedIPs(siteId int) []string {
+	var owned []string
+	for cidr, owner := range pm.allowedIPOwners {
+		if owner == siteId {
+			owned = append(owned, cidr)
+		}
+	}
+	return owned
+}
+
+// addAllowedIp adds an IP (subnet) to the allowed IPs list of a peer
+// and updates WireGuard configuration if this peer owns the IP.
+// Must be called with lock held.
+func (pm *PeerManager) addAllowedIp(siteId int, ip string) error {
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	// Check if IP already exists in AllowedIps
+	for _, allowedIp := range peer.AllowedIps {
+		if allowedIp == ip {
+			return nil // Already exists
+		}
+	}
+
+	// Register our claim to this IP
+	pm.claimAllowedIP(siteId, ip)
+
+	peer.AllowedIps = append(peer.AllowedIps, ip)
+	pm.peers[siteId] = peer
+
+	// Only update WireGuard if we own this IP
+	if pm.allowedIPOwners[ip] == siteId {
+		if err := AddAllowedIP(pm.device, peer.PublicKey, ip); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// removeAllowedIp removes an IP (subnet) from the allowed IPs list of a peer
+// and updates WireGuard configuration. If this peer owned the IP, it promotes
+// another peer that also claims this IP. Must be called with lock held.
+func (pm *PeerManager) removeAllowedIp(siteId int, cidr string) error {
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	found := false
+
+	// Remove from AllowedIps
+	newAllowedIps := make([]string, 0, len(peer.AllowedIps))
+	for _, allowedIp := range peer.AllowedIps {
+		if allowedIp == cidr {
+			found = true
+			continue
+		}
+		newAllowedIps = append(newAllowedIps, allowedIp)
+	}
+
+	if !found {
+		return nil // Not found
+	}
+
+	peer.AllowedIps = newAllowedIps
+	pm.peers[siteId] = peer
+
+	// Release our claim and check if we need to promote another peer
+	newOwner, promoted := pm.releaseAllowedIP(siteId, cidr)
+
+	// Build the list of IPs this peer currently owns for the replace operation
+	ownedIPs := pm.getOwnedAllowedIPs(siteId)
+	// Also include the server IP which is always owned
+	serverIP := strings.Split(peer.ServerIP, "/")[0] + "/32"
+	hasServerIP := false
+	for _, ip := range ownedIPs {
+		if ip == serverIP {
+			hasServerIP = true
+			break
+		}
+	}
+	if !hasServerIP {
+		ownedIPs = append([]string{serverIP}, ownedIPs...)
+	}
+
+	// Update WireGuard for this peer using replace_allowed_ips
+	if err := RemoveAllowedIP(pm.device, peer.PublicKey, ownedIPs); err != nil {
+		return err
+	}
+
+	// If another peer was promoted to owner, add the IP to their WireGuard config
+	if promoted && newOwner >= 0 {
+		if newOwnerPeer, exists := pm.peers[newOwner]; exists {
+			if err := AddAllowedIP(pm.device, newOwnerPeer.PublicKey, cidr); err != nil {
+				logger.Error("Failed to promote peer %d for IP %s: %v", newOwner, cidr, err)
+			} else {
+				logger.Info("Promoted peer %d to owner of IP %s", newOwner, cidr)
+			}
+		}
+	}
+
+	return nil
+}
+
+// AddRemoteSubnet adds an IP (subnet) to the allowed IPs list of a peer
+func (pm *PeerManager) AddRemoteSubnet(siteId int, cidr string) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	// Check if IP already exists in RemoteSubnets
+	for _, subnet := range peer.RemoteSubnets {
+		if subnet == cidr {
+			return nil // Already exists
+		}
+	}
+
+	peer.RemoteSubnets = append(peer.RemoteSubnets, cidr)
+	pm.peers[siteId] = peer // Save before calling addAllowedIp which reads from pm.peers
+
+	// Add to allowed IPs
+	if err := pm.addAllowedIp(siteId, cidr); err != nil {
+		return err
+	}
+
+	// Add route
+	if err := network.AddRoutes([]string{cidr}, pm.interfaceName); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// RemoveRemoteSubnet removes an IP (subnet) from the allowed IPs list of a peer
+func (pm *PeerManager) RemoveRemoteSubnet(siteId int, ip string) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	found := false
+
+	// Remove from RemoteSubnets
+	newSubnets := make([]string, 0, len(peer.RemoteSubnets))
+	for _, subnet := range peer.RemoteSubnets {
+		if subnet == ip {
+			found = true
+			continue
+		}
+		newSubnets = append(newSubnets, subnet)
+	}
+
+	if !found {
+		return nil // Not found
+	}
+
+	peer.RemoteSubnets = newSubnets
+	pm.peers[siteId] = peer // Save before calling removeAllowedIp which reads from pm.peers
+
+	// Remove from allowed IPs (this also handles promotion of other peers)
+	if err := pm.removeAllowedIp(siteId, ip); err != nil {
+		return err
+	}
+
+	// Check if any other peer still has this subnet before removing the route
+	subnetStillInUse := false
+	for otherSiteId, otherPeer := range pm.peers {
+		if otherSiteId == siteId {
+			continue // Skip the current peer (already updated above)
+		}
+		for _, subnet := range otherPeer.RemoteSubnets {
+			if subnet == ip {
+				subnetStillInUse = true
+				break
+			}
+		}
+		if subnetStillInUse {
+			break
+		}
+	}
+
+	// Only remove route if no other peer needs it
+	if !subnetStillInUse {
+		if err := network.RemoveRoutes([]string{ip}); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// AddAlias adds an alias to a peer
+func (pm *PeerManager) AddAlias(siteId int, alias Alias) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	peer.Aliases = append(peer.Aliases, alias)
+	pm.peers[siteId] = peer
+
+	address := net.ParseIP(alias.AliasAddress)
+	if address != nil {
+		pm.dnsProxy.AddDNSRecord(alias.Alias, address)
+	}
+
+	// Add an allowed IP for the alias
+	if err := pm.addAllowedIp(siteId, alias.AliasAddress+"/32"); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// RemoveAlias removes an alias from a peer
+func (pm *PeerManager) RemoveAlias(siteId int, aliasName string) error {
+	pm.mu.Lock()
+	defer pm.mu.Unlock()
+
+	peer, exists := pm.peers[siteId]
+	if !exists {
+		return fmt.Errorf("peer with site ID %d not found", siteId)
+	}
+
+	var aliasToRemove *Alias
+	newAliases := make([]Alias, 0, len(peer.Aliases))
+	for _, a := range peer.Aliases {
+		if a.Alias == aliasName {
+			aliasToRemove = &a
+			continue
+		}
+		newAliases = append(newAliases, a)
+	}
+
+	if aliasToRemove != nil {
+		address := net.ParseIP(aliasToRemove.AliasAddress)
+		if address != nil {
+			pm.dnsProxy.RemoveDNSRecord(aliasName, address)
+		}
+	}
+
+	peer.Aliases = newAliases
+	pm.peers[siteId] = peer
+
+	// Check if any other alias is still using this IP address before removing from allowed IPs
+	ipStillInUse := false
+	aliasIP := aliasToRemove.AliasAddress + "/32"
+	for _, a := range newAliases {
+		if a.AliasAddress+"/32" == aliasIP {
+			ipStillInUse = true
+			break
+		}
+	}
+
+	// Only remove the allowed IP if no other alias is using it
+	if !ipStillInUse {
+		if err := pm.removeAllowedIp(siteId, aliasIP); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// RelayPeer handles failover to the relay server when a peer is disconnected
+func (pm *PeerManager) RelayPeer(siteId int, relayEndpoint string, relayPort uint16) {
+	pm.mu.Lock()
+	peer, exists := pm.peers[siteId]
+	if exists {
+		// Store the relay endpoint
+		peer.RelayEndpoint = relayEndpoint
+		pm.peers[siteId] = peer
+	}
+	pm.mu.Unlock()
+
+	if !exists {
+		logger.Error("Cannot handle failover: peer with site ID %d not found", siteId)
+		return
+	}
+
+	// Check for IPv6 and format the endpoint correctly
+	formattedEndpoint := relayEndpoint
+	if strings.Contains(relayEndpoint, ":") {
+		formattedEndpoint = fmt.Sprintf("[%s]", relayEndpoint)
+	}
+
+	if relayPort == 0 {
+		relayPort = 21820 // fall back to 21820 for backward compatibility
+	}
+
+	// Update only the endpoint for this peer (update_only preserves other settings)
+	wgConfig := fmt.Sprintf(`public_key=%s
+update_only=true
+endpoint=%s:%d`, util.FixKey(peer.PublicKey), formattedEndpoint, relayPort)
+
+	err := pm.device.IpcSet(wgConfig)
+	if err != nil {
+		logger.Error("Failed to configure WireGuard device: %v\n", err)
+		return
+	}
+
+	// Mark the peer as relayed in the monitor
+	if pm.peerMonitor != nil {
+		pm.peerMonitor.MarkPeerRelayed(siteId, true)
+	}
+
+	logger.Info("Adjusted peer %d to point to relay!\n", siteId)
+}
+
+// performRapidInitialTest performs a rapid holepunch test for a newly added peer.
+// If the test fails, it immediately requests relay to minimize connection delay.
+// This runs in a goroutine to avoid blocking AddPeer.
+func (pm *PeerManager) performRapidInitialTest(siteId int, endpoint string) {
+	if pm.peerMonitor == nil {
+		return
+	}
+
+	// Perform rapid test - this takes ~1-2 seconds max
+	holepunchViable := pm.peerMonitor.RapidTestPeer(siteId, endpoint)
+
+	if !holepunchViable {
+		// Holepunch failed rapid test, request relay immediately
+		logger.Info("Rapid test failed for site %d, requesting relay", siteId)
+		if err := pm.peerMonitor.RequestRelay(siteId); err != nil {
+			logger.Error("Failed to request relay for site %d: %v", siteId, err)
+		}
+	} else {
+		logger.Info("Rapid test passed for site %d, using direct connection", siteId)
+	}
+}
+
+// Start starts the peer monitor
+func (pm *PeerManager) Start() {
+	if pm.peerMonitor != nil {
+		pm.peerMonitor.Start()
+	}
+}
+
+// Stop stops the peer monitor
+func (pm *PeerManager) Stop() {
+	if pm.peerMonitor != nil {
+		pm.peerMonitor.Stop()
+	}
+}
+
+// Close stops the peer monitor and cleans up resources
+func (pm *PeerManager) Close() {
+	if pm.peerMonitor != nil {
+		pm.peerMonitor.Close()
+		pm.peerMonitor = nil
+	}
+}
+
+// MarkPeerRelayed marks a peer as currently using relay
+func (pm *PeerManager) MarkPeerRelayed(siteID int, relayed bool) {
+	pm.mu.Lock()
+	if peer, exists := pm.peers[siteID]; exists {
+		if relayed {
+			// We're being relayed, store the current endpoint as the original
+			// (RelayEndpoint is set by HandleFailover)
+		} else {
+			// Clear relay endpoint when switching back to direct
+			peer.RelayEndpoint = ""
+			pm.peers[siteID] = peer
+		}
+	}
+	pm.mu.Unlock()
+
+	if pm.peerMonitor != nil {
+		pm.peerMonitor.MarkPeerRelayed(siteID, relayed)
+	}
+}
+
+// UnRelayPeer switches a peer from relay back to direct connection
+func (pm *PeerManager) UnRelayPeer(siteId int, endpoint string) error {
+	pm.mu.Lock()
+	peer, exists := pm.peers[siteId]
+	if exists {
+		// Store the relay endpoint
+		peer.Endpoint = endpoint
+		pm.peers[siteId] = peer
+	}
+	pm.mu.Unlock()
+
+	if !exists {
+		logger.Error("Cannot handle failover: peer with site ID %d not found", siteId)
+		return nil
+	}
+
+	// Update WireGuard to use the direct endpoint
+	wgConfig := fmt.Sprintf(`public_key=%s
+update_only=true
+endpoint=%s`, util.FixKey(peer.PublicKey), endpoint)
+
+	err := pm.device.IpcSet(wgConfig)
+	if err != nil {
+		logger.Error("Failed to switch peer %d to direct connection: %v", siteId, err)
+		return err
+	}
+
+	// Mark as not relayed in monitor
+	if pm.peerMonitor != nil {
+		pm.peerMonitor.MarkPeerRelayed(siteId, false)
+	}
+
+	logger.Info("Switched peer %d back to direct connection at %s", siteId, endpoint)
+	return nil
+}

peers/monitor/monitor.go

@@ -0,0 +1,930 @@
+package monitor
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"net/netip"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/bind"
+	"github.com/fosrl/newt/holepunch"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/util"
+	"github.com/fosrl/olm/api"
+	middleDevice "github.com/fosrl/olm/device"
+	"github.com/fosrl/olm/websocket"
+	"gvisor.dev/gvisor/pkg/buffer"
+	"gvisor.dev/gvisor/pkg/tcpip"
+	"gvisor.dev/gvisor/pkg/tcpip/adapters/gonet"
+	"gvisor.dev/gvisor/pkg/tcpip/header"
+	"gvisor.dev/gvisor/pkg/tcpip/link/channel"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv6"
+	"gvisor.dev/gvisor/pkg/tcpip/stack"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/udp"
+)
+
+// PeerMonitor handles monitoring the connection status to multiple WireGuard peers
+type PeerMonitor struct {
+	monitors    map[int]*Client
+	mutex       sync.Mutex
+	running     bool
+	interval    time.Duration
+	timeout     time.Duration
+	maxAttempts int
+	wsClient    *websocket.Client
+
+	// Netstack fields
+	middleDev   *middleDevice.MiddleDevice
+	localIP     string
+	stack       *stack.Stack
+	ep          *channel.Endpoint
+	activePorts map[uint16]bool
+	portsLock   sync.Mutex
+	nsCtx       context.Context
+	nsCancel    context.CancelFunc
+	nsWg        sync.WaitGroup
+
+	// Holepunch testing fields
+	sharedBind         *bind.SharedBind
+	holepunchTester    *holepunch.HolepunchTester
+	holepunchInterval  time.Duration
+	holepunchTimeout   time.Duration
+	holepunchEndpoints map[int]string // siteID -> endpoint for holepunch testing
+	holepunchStatus    map[int]bool   // siteID -> connected status
+	holepunchStopChan  chan struct{}
+
+	// Relay tracking fields
+	relayedPeers         map[int]bool // siteID -> whether the peer is currently relayed
+	holepunchMaxAttempts int          // max consecutive failures before triggering relay
+	holepunchFailures    map[int]int  // siteID -> consecutive failure count
+
+	// Rapid initial test fields
+	rapidTestInterval    time.Duration // interval between rapid test attempts
+	rapidTestTimeout     time.Duration // timeout for each rapid test attempt
+	rapidTestMaxAttempts int           // max attempts during rapid test phase
+
+	// API server for status updates
+	apiServer *api.API
+
+	// WG connection status tracking
+	wgConnectionStatus map[int]bool // siteID -> WG connected status
+}
+
+// NewPeerMonitor creates a new peer monitor with the given callback
+func NewPeerMonitor(wsClient *websocket.Client, middleDev *middleDevice.MiddleDevice, localIP string, sharedBind *bind.SharedBind, apiServer *api.API) *PeerMonitor {
+	ctx, cancel := context.WithCancel(context.Background())
+	pm := &PeerMonitor{
+		monitors:             make(map[int]*Client),
+		interval:             2 * time.Second, // Default check interval (faster)
+		timeout:              3 * time.Second,
+		maxAttempts:          3,
+		wsClient:             wsClient,
+		middleDev:            middleDev,
+		localIP:              localIP,
+		activePorts:          make(map[uint16]bool),
+		nsCtx:                ctx,
+		nsCancel:             cancel,
+		sharedBind:           sharedBind,
+		holepunchInterval:    2 * time.Second, // Check holepunch every 2 seconds
+		holepunchTimeout:     2 * time.Second, // Faster timeout
+		holepunchEndpoints:   make(map[int]string),
+		holepunchStatus:      make(map[int]bool),
+		relayedPeers:         make(map[int]bool),
+		holepunchMaxAttempts: 3, // Trigger relay after 3 consecutive failures
+		holepunchFailures:    make(map[int]int),
+		// Rapid initial test settings: complete within ~1.5 seconds
+		rapidTestInterval:    200 * time.Millisecond, // 200ms between attempts
+		rapidTestTimeout:     400 * time.Millisecond, // 400ms timeout per attempt
+		rapidTestMaxAttempts: 5,                      // 5 attempts = ~1-1.5 seconds total
+		apiServer:            apiServer,
+		wgConnectionStatus:   make(map[int]bool),
+	}
+
+	if err := pm.initNetstack(); err != nil {
+		logger.Error("Failed to initialize netstack for peer monitor: %v", err)
+	}
+
+	// Initialize holepunch tester if sharedBind is available
+	if sharedBind != nil {
+		pm.holepunchTester = holepunch.NewHolepunchTester(sharedBind)
+	}
+
+	return pm
+}
+
+// SetInterval changes how frequently peers are checked
+func (pm *PeerMonitor) SetInterval(interval time.Duration) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	pm.interval = interval
+
+	// Update interval for all existing monitors
+	for _, client := range pm.monitors {
+		client.SetPacketInterval(interval)
+	}
+}
+
+// SetTimeout changes the timeout for waiting for responses
+func (pm *PeerMonitor) SetTimeout(timeout time.Duration) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	pm.timeout = timeout
+
+	// Update timeout for all existing monitors
+	for _, client := range pm.monitors {
+		client.SetTimeout(timeout)
+	}
+}
+
+// SetMaxAttempts changes the maximum number of attempts for TestConnection
+func (pm *PeerMonitor) SetMaxAttempts(attempts int) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	pm.maxAttempts = attempts
+
+	// Update max attempts for all existing monitors
+	for _, client := range pm.monitors {
+		client.SetMaxAttempts(attempts)
+	}
+}
+
+// AddPeer adds a new peer to monitor
+func (pm *PeerMonitor) AddPeer(siteID int, endpoint string, holepunchEndpoint string) error {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	if _, exists := pm.monitors[siteID]; exists {
+		return nil // Already monitoring
+	}
+
+	// Use our custom dialer that uses netstack
+	client, err := NewClient(endpoint, pm.dial)
+	if err != nil {
+		return err
+	}
+
+	client.SetPacketInterval(pm.interval)
+	client.SetTimeout(pm.timeout)
+	client.SetMaxAttempts(pm.maxAttempts)
+
+	pm.monitors[siteID] = client
+
+	pm.holepunchEndpoints[siteID] = holepunchEndpoint
+	pm.holepunchStatus[siteID] = false // Initially unknown/disconnected
+
+	if pm.running {
+		if err := client.StartMonitor(func(status ConnectionStatus) {
+			pm.handleConnectionStatusChange(siteID, status)
+		}); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// update holepunch endpoint for a peer
+func (pm *PeerMonitor) UpdateHolepunchEndpoint(siteID int, endpoint string) {
+	// Short delay to allow WireGuard peer reconfiguration to complete
+	// The NAT mapping refresh is handled separately by TriggerHolePunch in olm.go
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+	pm.holepunchEndpoints[siteID] = endpoint
+	logger.Debug("Updated holepunch endpoint for site %d to %s", siteID, endpoint)
+}
+
+// RapidTestPeer performs a rapid connectivity test for a newly added peer.
+// This is designed to quickly determine if holepunch is viable within ~1-2 seconds.
+// Returns true if the connection is viable (holepunch works), false if it should relay.
+func (pm *PeerMonitor) RapidTestPeer(siteID int, endpoint string) bool {
+	if pm.holepunchTester == nil {
+		logger.Warn("Cannot perform rapid test: holepunch tester not initialized")
+		return false
+	}
+
+	pm.mutex.Lock()
+	interval := pm.rapidTestInterval
+	timeout := pm.rapidTestTimeout
+	maxAttempts := pm.rapidTestMaxAttempts
+	pm.mutex.Unlock()
+
+	logger.Info("Starting rapid holepunch test for site %d at %s (max %d attempts, %v timeout each)",
+		siteID, endpoint, maxAttempts, timeout)
+
+	for attempt := 1; attempt <= maxAttempts; attempt++ {
+		result := pm.holepunchTester.TestEndpoint(endpoint, timeout)
+
+		if result.Success {
+			logger.Info("Rapid test: site %d holepunch SUCCEEDED on attempt %d (RTT: %v)",
+				siteID, attempt, result.RTT)
+
+			// Update status
+			pm.mutex.Lock()
+			pm.holepunchStatus[siteID] = true
+			pm.holepunchFailures[siteID] = 0
+			pm.mutex.Unlock()
+
+			return true
+		}
+
+		if attempt < maxAttempts {
+			time.Sleep(interval)
+		}
+	}
+
+	logger.Warn("Rapid test: site %d holepunch FAILED after %d attempts, will relay",
+		siteID, maxAttempts)
+
+	// Update status to reflect failure
+	pm.mutex.Lock()
+	pm.holepunchStatus[siteID] = false
+	pm.holepunchFailures[siteID] = maxAttempts
+	pm.mutex.Unlock()
+
+	return false
+}
+
+// UpdatePeerEndpoint updates the monitor endpoint for a peer
+func (pm *PeerMonitor) UpdatePeerEndpoint(siteID int, monitorPeer string) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	client, exists := pm.monitors[siteID]
+	if !exists {
+		logger.Warn("Cannot update endpoint: peer %d not found in monitor", siteID)
+		return
+	}
+
+	// Update the client's server address
+	client.UpdateServerAddr(monitorPeer)
+
+	logger.Info("Updated monitor endpoint for site %d to %s", siteID, monitorPeer)
+}
+
+// removePeerUnlocked stops monitoring a peer and removes it from the monitor
+// This function assumes the mutex is already held by the caller
+func (pm *PeerMonitor) removePeerUnlocked(siteID int) {
+	client, exists := pm.monitors[siteID]
+	if !exists {
+		return
+	}
+
+	client.StopMonitor()
+	client.Close()
+	delete(pm.monitors, siteID)
+}
+
+// RemovePeer stops monitoring a peer and removes it from the monitor
+func (pm *PeerMonitor) RemovePeer(siteID int) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	// remove the holepunch endpoint info
+	delete(pm.holepunchEndpoints, siteID)
+	delete(pm.holepunchStatus, siteID)
+	delete(pm.relayedPeers, siteID)
+	delete(pm.holepunchFailures, siteID)
+
+	pm.removePeerUnlocked(siteID)
+}
+
+func (pm *PeerMonitor) RemoveHolepunchEndpoint(siteID int) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+	delete(pm.holepunchEndpoints, siteID)
+}
+
+// Start begins monitoring all peers
+func (pm *PeerMonitor) Start() {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	if pm.running {
+		return // Already running
+	}
+
+	pm.running = true
+
+	// Start monitoring all peers
+	for siteID, client := range pm.monitors {
+		siteIDCopy := siteID // Create a copy for the closure
+		err := client.StartMonitor(func(status ConnectionStatus) {
+			pm.handleConnectionStatusChange(siteIDCopy, status)
+		})
+		if err != nil {
+			logger.Error("Failed to start monitoring peer %d: %v\n", siteID, err)
+			continue
+		}
+		logger.Info("Started monitoring peer %d\n", siteID)
+	}
+
+	pm.startHolepunchMonitor()
+}
+
+// handleConnectionStatusChange is called when a peer's connection status changes
+func (pm *PeerMonitor) handleConnectionStatusChange(siteID int, status ConnectionStatus) {
+	pm.mutex.Lock()
+	previousStatus, exists := pm.wgConnectionStatus[siteID]
+	pm.wgConnectionStatus[siteID] = status.Connected
+	isRelayed := pm.relayedPeers[siteID]
+	endpoint := pm.holepunchEndpoints[siteID]
+	pm.mutex.Unlock()
+
+	// Log status changes
+	if !exists || previousStatus != status.Connected {
+		if status.Connected {
+			logger.Info("WireGuard connection to site %d is CONNECTED (RTT: %v)", siteID, status.RTT)
+		} else {
+			logger.Warn("WireGuard connection to site %d is DISCONNECTED", siteID)
+		}
+	}
+
+	// Update API with connection status
+	if pm.apiServer != nil {
+		pm.apiServer.UpdatePeerStatus(siteID, status.Connected, status.RTT, endpoint, isRelayed)
+	}
+}
+
+// sendRelay sends a relay message to the server
+func (pm *PeerMonitor) sendRelay(siteID int) error {
+	if pm.wsClient == nil {
+		return fmt.Errorf("websocket client is nil")
+	}
+
+	err := pm.wsClient.SendMessage("olm/wg/relay", map[string]interface{}{
+		"siteId": siteID,
+	})
+	if err != nil {
+		logger.Error("Failed to send registration message: %v", err)
+		return err
+	}
+	logger.Info("Sent relay message")
+	return nil
+}
+
+// RequestRelay is a public method to request relay for a peer.
+// This is used when rapid initial testing determines holepunch is not viable.
+func (pm *PeerMonitor) RequestRelay(siteID int) error {
+	return pm.sendRelay(siteID)
+}
+
+// sendUnRelay sends an unrelay message to the server
+func (pm *PeerMonitor) sendUnRelay(siteID int) error {
+	if pm.wsClient == nil {
+		return fmt.Errorf("websocket client is nil")
+	}
+
+	err := pm.wsClient.SendMessage("olm/wg/unrelay", map[string]interface{}{
+		"siteId": siteID,
+	})
+	if err != nil {
+		logger.Error("Failed to send registration message: %v", err)
+		return err
+	}
+	logger.Info("Sent unrelay message")
+	return nil
+}
+
+// Stop stops monitoring all peers
+func (pm *PeerMonitor) Stop() {
+	// Stop holepunch monitor first (outside of mutex to avoid deadlock)
+	pm.stopHolepunchMonitor()
+
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	if !pm.running {
+		return
+	}
+
+	pm.running = false
+
+	// Stop all monitors
+	for _, client := range pm.monitors {
+		client.StopMonitor()
+	}
+}
+
+// MarkPeerRelayed marks a peer as currently using relay
+func (pm *PeerMonitor) MarkPeerRelayed(siteID int, relayed bool) {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+	pm.relayedPeers[siteID] = relayed
+	if relayed {
+		// Reset failure count when marked as relayed
+		pm.holepunchFailures[siteID] = 0
+	}
+}
+
+// IsPeerRelayed returns whether a peer is currently using relay
+func (pm *PeerMonitor) IsPeerRelayed(siteID int) bool {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+	return pm.relayedPeers[siteID]
+}
+
+// startHolepunchMonitor starts the holepunch connection monitoring
+// Note: This function assumes the mutex is already held by the caller (called from Start())
+func (pm *PeerMonitor) startHolepunchMonitor() error {
+	if pm.holepunchTester == nil {
+		return fmt.Errorf("holepunch tester not initialized (sharedBind not provided)")
+	}
+
+	if pm.holepunchStopChan != nil {
+		return fmt.Errorf("holepunch monitor already running")
+	}
+
+	if err := pm.holepunchTester.Start(); err != nil {
+		return fmt.Errorf("failed to start holepunch tester: %w", err)
+	}
+
+	pm.holepunchStopChan = make(chan struct{})
+
+	go pm.runHolepunchMonitor()
+
+	logger.Info("Started holepunch connection monitor")
+	return nil
+}
+
+// stopHolepunchMonitor stops the holepunch connection monitoring
+func (pm *PeerMonitor) stopHolepunchMonitor() {
+	pm.mutex.Lock()
+	stopChan := pm.holepunchStopChan
+	pm.holepunchStopChan = nil
+	pm.mutex.Unlock()
+
+	if stopChan != nil {
+		close(stopChan)
+	}
+
+	if pm.holepunchTester != nil {
+		pm.holepunchTester.Stop()
+	}
+
+	logger.Info("Stopped holepunch connection monitor")
+}
+
+// runHolepunchMonitor runs the holepunch monitoring loop
+func (pm *PeerMonitor) runHolepunchMonitor() {
+	ticker := time.NewTicker(pm.holepunchInterval)
+	defer ticker.Stop()
+
+	// Do initial check immediately
+	pm.checkHolepunchEndpoints()
+
+	for {
+		select {
+		case <-pm.holepunchStopChan:
+			return
+		case <-ticker.C:
+			pm.checkHolepunchEndpoints()
+		}
+	}
+}
+
+// checkHolepunchEndpoints tests all holepunch endpoints
+func (pm *PeerMonitor) checkHolepunchEndpoints() {
+	pm.mutex.Lock()
+	// Check if we're still running before doing any work
+	if !pm.running {
+		pm.mutex.Unlock()
+		return
+	}
+	endpoints := make(map[int]string, len(pm.holepunchEndpoints))
+	for siteID, endpoint := range pm.holepunchEndpoints {
+		endpoints[siteID] = endpoint
+	}
+	timeout := pm.holepunchTimeout
+	maxAttempts := pm.holepunchMaxAttempts
+	pm.mutex.Unlock()
+
+	for siteID, endpoint := range endpoints {
+		// logger.Debug("Testing holepunch endpoint for site %d: %s", siteID, endpoint)
+		result := pm.holepunchTester.TestEndpoint(endpoint, timeout)
+
+		pm.mutex.Lock()
+		// Check if peer was removed while we were testing
+		if _, stillExists := pm.holepunchEndpoints[siteID]; !stillExists {
+			pm.mutex.Unlock()
+			continue // Peer was removed, skip processing
+		}
+
+		previousStatus, exists := pm.holepunchStatus[siteID]
+		pm.holepunchStatus[siteID] = result.Success
+		isRelayed := pm.relayedPeers[siteID]
+
+		// Track consecutive failures for relay triggering
+		if result.Success {
+			pm.holepunchFailures[siteID] = 0
+		} else {
+			pm.holepunchFailures[siteID]++
+		}
+		failureCount := pm.holepunchFailures[siteID]
+		pm.mutex.Unlock()
+
+		// Log status changes
+		if !exists || previousStatus != result.Success {
+			if result.Success {
+				logger.Info("Holepunch to site %d (%s) is CONNECTED (RTT: %v)", siteID, endpoint, result.RTT)
+			} else {
+				if result.Error != nil {
+					logger.Warn("Holepunch to site %d (%s) is DISCONNECTED: %v", siteID, endpoint, result.Error)
+				} else {
+					logger.Warn("Holepunch to site %d (%s) is DISCONNECTED", siteID, endpoint)
+				}
+			}
+		}
+
+		// Update API with holepunch status
+		if pm.apiServer != nil {
+			// Update holepunch connection status
+			pm.apiServer.UpdatePeerHolepunchStatus(siteID, result.Success)
+
+			// Get the current WG connection status for this peer
+			pm.mutex.Lock()
+			wgConnected := pm.wgConnectionStatus[siteID]
+			pm.mutex.Unlock()
+
+			// Update API - use holepunch endpoint and relay status
+			pm.apiServer.UpdatePeerStatus(siteID, wgConnected, result.RTT, endpoint, isRelayed)
+		}
+
+		// Handle relay logic based on holepunch status
+		// Check if we're still running before sending relay messages
+		pm.mutex.Lock()
+		stillRunning := pm.running
+		pm.mutex.Unlock()
+
+		if !stillRunning {
+			return // Stop processing if shutdown is in progress
+		}
+
+		if !result.Success && !isRelayed && failureCount >= maxAttempts {
+			// Holepunch failed and we're not relayed - trigger relay
+			logger.Info("Holepunch to site %d failed %d times, triggering relay", siteID, failureCount)
+			if pm.wsClient != nil {
+				pm.sendRelay(siteID)
+			}
+		} else if result.Success && isRelayed {
+			// Holepunch succeeded and we ARE relayed - switch back to direct
+			logger.Info("Holepunch to site %d succeeded while relayed, switching to direct connection", siteID)
+			if pm.wsClient != nil {
+				pm.sendUnRelay(siteID)
+			}
+		}
+	}
+}
+
+// GetHolepunchStatus returns the current holepunch status for all endpoints
+func (pm *PeerMonitor) GetHolepunchStatus() map[int]bool {
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	status := make(map[int]bool, len(pm.holepunchStatus))
+	for siteID, connected := range pm.holepunchStatus {
+		status[siteID] = connected
+	}
+	return status
+}
+
+// Close stops monitoring and cleans up resources
+func (pm *PeerMonitor) Close() {
+	// Stop holepunch monitor first (outside of mutex to avoid deadlock)
+	pm.stopHolepunchMonitor()
+
+	pm.mutex.Lock()
+	defer pm.mutex.Unlock()
+
+	logger.Debug("PeerMonitor: Starting cleanup")
+
+	// Stop and close all clients first
+	for siteID, client := range pm.monitors {
+		logger.Debug("PeerMonitor: Stopping client for site %d", siteID)
+		client.StopMonitor()
+		client.Close()
+		delete(pm.monitors, siteID)
+	}
+
+	pm.running = false
+
+	// Clean up netstack resources
+	logger.Debug("PeerMonitor: Cancelling netstack context")
+	if pm.nsCancel != nil {
+		pm.nsCancel() // Signal goroutines to stop
+	}
+
+	// Close the channel endpoint to unblock any pending reads
+	logger.Debug("PeerMonitor: Closing endpoint")
+	if pm.ep != nil {
+		pm.ep.Close()
+	}
+
+	// Wait for packet sender goroutine to finish with timeout
+	logger.Debug("PeerMonitor: Waiting for goroutines to finish")
+	done := make(chan struct{})
+	go func() {
+		pm.nsWg.Wait()
+		close(done)
+	}()
+
+	select {
+	case <-done:
+		logger.Debug("PeerMonitor: Goroutines finished cleanly")
+	case <-time.After(2 * time.Second):
+		logger.Warn("PeerMonitor: Timeout waiting for goroutines to finish, proceeding anyway")
+	}
+
+	// Destroy the stack last, after all goroutines are done
+	logger.Debug("PeerMonitor: Destroying stack")
+	if pm.stack != nil {
+		pm.stack.Destroy()
+		pm.stack = nil
+	}
+
+	logger.Debug("PeerMonitor: Cleanup complete")
+}
+
+// TestPeer tests connectivity to a specific peer
+func (pm *PeerMonitor) TestPeer(siteID int) (bool, time.Duration, error) {
+	pm.mutex.Lock()
+	client, exists := pm.monitors[siteID]
+	pm.mutex.Unlock()
+
+	if !exists {
+		return false, 0, fmt.Errorf("peer with siteID %d not found", siteID)
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), pm.timeout*time.Duration(pm.maxAttempts))
+	defer cancel()
+
+	connected, rtt := client.TestConnection(ctx)
+	return connected, rtt, nil
+}
+
+// TestAllPeers tests connectivity to all peers
+func (pm *PeerMonitor) TestAllPeers() map[int]struct {
+	Connected bool
+	RTT       time.Duration
+} {
+	pm.mutex.Lock()
+	peers := make(map[int]*Client, len(pm.monitors))
+	for siteID, client := range pm.monitors {
+		peers[siteID] = client
+	}
+	pm.mutex.Unlock()
+
+	results := make(map[int]struct {
+		Connected bool
+		RTT       time.Duration
+	})
+	for siteID, client := range peers {
+		ctx, cancel := context.WithTimeout(context.Background(), pm.timeout*time.Duration(pm.maxAttempts))
+		connected, rtt := client.TestConnection(ctx)
+		cancel()
+
+		results[siteID] = struct {
+			Connected bool
+			RTT       time.Duration
+		}{
+			Connected: connected,
+			RTT:       rtt,
+		}
+	}
+
+	return results
+}
+
+// initNetstack initializes the gvisor netstack
+func (pm *PeerMonitor) initNetstack() error {
+	if pm.localIP == "" {
+		return fmt.Errorf("local IP not provided")
+	}
+
+	addr, err := netip.ParseAddr(pm.localIP)
+	if err != nil {
+		return fmt.Errorf("invalid local IP: %v", err)
+	}
+
+	// Create gvisor netstack
+	stackOpts := stack.Options{
+		NetworkProtocols:   []stack.NetworkProtocolFactory{ipv4.NewProtocol, ipv6.NewProtocol},
+		TransportProtocols: []stack.TransportProtocolFactory{udp.NewProtocol},
+		HandleLocal:        true,
+	}
+
+	pm.ep = channel.New(256, 1420, "") // MTU 1420 (standard WG)
+	pm.stack = stack.New(stackOpts)
+
+	// Create NIC
+	if err := pm.stack.CreateNIC(1, pm.ep); err != nil {
+		return fmt.Errorf("failed to create NIC: %v", err)
+	}
+
+	// Add IP address
+	ipBytes := addr.As4()
+	protoAddr := tcpip.ProtocolAddress{
+		Protocol:          ipv4.ProtocolNumber,
+		AddressWithPrefix: tcpip.AddrFrom4(ipBytes).WithPrefix(),
+	}
+
+	if err := pm.stack.AddProtocolAddress(1, protoAddr, stack.AddressProperties{}); err != nil {
+		return fmt.Errorf("failed to add protocol address: %v", err)
+	}
+
+	// Add default route
+	pm.stack.AddRoute(tcpip.Route{
+		Destination: header.IPv4EmptySubnet,
+		NIC:         1,
+	})
+
+	// Register filter rule on MiddleDevice
+	// We want to intercept packets destined to our local IP
+	// But ONLY if they are for ports we are listening on
+	pm.middleDev.AddRule(addr, pm.handlePacket)
+
+	// Start packet sender (Stack -> WG)
+	pm.nsWg.Add(1)
+	go pm.runPacketSender()
+
+	return nil
+}
+
+// handlePacket is called by MiddleDevice when a packet arrives for our IP
+func (pm *PeerMonitor) handlePacket(packet []byte) bool {
+	// Check if it's UDP
+	proto, ok := util.GetProtocol(packet)
+	if !ok || proto != 17 { // UDP
+		return false
+	}
+
+	// Check destination port
+	port, ok := util.GetDestPort(packet)
+	if !ok {
+		return false
+	}
+
+	// Check if we are listening on this port
+	pm.portsLock.Lock()
+	active := pm.activePorts[uint16(port)]
+	pm.portsLock.Unlock()
+
+	if !active {
+		return false
+	}
+
+	// Inject into netstack
+	version := packet[0] >> 4
+	pkb := stack.NewPacketBuffer(stack.PacketBufferOptions{
+		Payload: buffer.MakeWithData(packet),
+	})
+
+	switch version {
+	case 4:
+		pm.ep.InjectInbound(ipv4.ProtocolNumber, pkb)
+	case 6:
+		pm.ep.InjectInbound(ipv6.ProtocolNumber, pkb)
+	default:
+		pkb.DecRef()
+		return false
+	}
+
+	pkb.DecRef()
+	return true // Handled
+}
+
+// runPacketSender reads packets from netstack and injects them into WireGuard
+func (pm *PeerMonitor) runPacketSender() {
+	defer pm.nsWg.Done()
+	logger.Debug("PeerMonitor: Packet sender goroutine started")
+
+	// Use a ticker to periodically check for packets without blocking indefinitely
+	ticker := time.NewTicker(10 * time.Millisecond)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-pm.nsCtx.Done():
+			logger.Debug("PeerMonitor: Packet sender context cancelled, draining packets")
+			// Drain any remaining packets before exiting
+			for {
+				pkt := pm.ep.Read()
+				if pkt == nil {
+					break
+				}
+				pkt.DecRef()
+			}
+			logger.Debug("PeerMonitor: Packet sender goroutine exiting")
+			return
+		case <-ticker.C:
+			// Try to read packets in batches
+			for i := 0; i < 10; i++ {
+				pkt := pm.ep.Read()
+				if pkt == nil {
+					break
+				}
+
+				// Extract packet data
+				slices := pkt.AsSlices()
+				if len(slices) > 0 {
+					var totalSize int
+					for _, slice := range slices {
+						totalSize += len(slice)
+					}
+
+					buf := make([]byte, totalSize)
+					pos := 0
+					for _, slice := range slices {
+						copy(buf[pos:], slice)
+						pos += len(slice)
+					}
+
+					// Inject into MiddleDevice (outbound to WG)
+					pm.middleDev.InjectOutbound(buf)
+				}
+
+				pkt.DecRef()
+			}
+		}
+	}
+}
+
+// dial creates a UDP connection using the netstack
+func (pm *PeerMonitor) dial(network, addr string) (net.Conn, error) {
+	if pm.stack == nil {
+		return nil, fmt.Errorf("netstack not initialized")
+	}
+
+	// Parse remote address
+	raddr, err := net.ResolveUDPAddr("udp", addr)
+	if err != nil {
+		return nil, err
+	}
+
+	// Parse local IP
+	localIP, err := netip.ParseAddr(pm.localIP)
+	if err != nil {
+		return nil, err
+	}
+	ipBytes := localIP.As4()
+
+	// Create UDP connection
+	// We bind to port 0 (ephemeral)
+	laddr := &tcpip.FullAddress{
+		NIC:  1,
+		Addr: tcpip.AddrFrom4(ipBytes),
+		Port: 0,
+	}
+
+	raddrTcpip := &tcpip.FullAddress{
+		NIC:  1,
+		Addr: tcpip.AddrFrom4([4]byte(raddr.IP.To4())),
+		Port: uint16(raddr.Port),
+	}
+
+	conn, err := gonet.DialUDP(pm.stack, laddr, raddrTcpip, ipv4.ProtocolNumber)
+	if err != nil {
+		return nil, err
+	}
+
+	// Get local port
+	localAddr := conn.LocalAddr().(*net.UDPAddr)
+	port := uint16(localAddr.Port)
+
+	// Register port
+	pm.portsLock.Lock()
+	pm.activePorts[port] = true
+	pm.portsLock.Unlock()
+
+	// Wrap connection to cleanup port on close
+	return &trackedConn{
+		Conn: conn,
+		pm:   pm,
+		port: port,
+	}, nil
+}
+
+func (pm *PeerMonitor) removePort(port uint16) {
+	pm.portsLock.Lock()
+	delete(pm.activePorts, port)
+	pm.portsLock.Unlock()
+}
+
+type trackedConn struct {
+	net.Conn
+	pm   *PeerMonitor
+	port uint16
+}
+
+func (c *trackedConn) Close() error {
+	c.pm.removePort(c.port)
+	if c.Conn != nil {
+		return c.Conn.Close()
+	}
+	return nil
+}

peers/monitor/wgtester.go

@@ -0,0 +1,281 @@
+package monitor
+
+import (
+	"context"
+	"encoding/binary"
+	"net"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+)
+
+const (
+	// Magic bytes to identify our packets
+	magicHeader uint32 = 0xDEADBEEF
+	// Request packet type
+	packetTypeRequest uint8 = 1
+	// Response packet type
+	packetTypeResponse uint8 = 2
+	// Packet format:
+	// - 4 bytes: magic header (0xDEADBEEF)
+	// - 1 byte: packet type (1 = request, 2 = response)
+	// - 8 bytes: timestamp (for round-trip timing)
+	packetSize = 13
+)
+
+// Client handles checking connectivity to a server
+type Client struct {
+	conn           net.Conn
+	serverAddr     string
+	monitorRunning bool
+	monitorLock    sync.Mutex
+	connLock       sync.Mutex // Protects connection operations
+	shutdownCh     chan struct{}
+	packetInterval time.Duration
+	timeout        time.Duration
+	maxAttempts    int
+	dialer         Dialer
+}
+
+// Dialer is a function that creates a connection
+type Dialer func(network, addr string) (net.Conn, error)
+
+// ConnectionStatus represents the current connection state
+type ConnectionStatus struct {
+	Connected bool
+	RTT       time.Duration
+}
+
+// NewClient creates a new connection test client
+func NewClient(serverAddr string, dialer Dialer) (*Client, error) {
+	return &Client{
+		serverAddr:     serverAddr,
+		shutdownCh:     make(chan struct{}),
+		packetInterval: 2 * time.Second,
+		timeout:        500 * time.Millisecond, // Timeout for individual packets
+		maxAttempts:    3,                      // Default max attempts
+		dialer:         dialer,
+	}, nil
+}
+
+// SetPacketInterval changes how frequently packets are sent in monitor mode
+func (c *Client) SetPacketInterval(interval time.Duration) {
+	c.packetInterval = interval
+}
+
+// SetTimeout changes the timeout for waiting for responses
+func (c *Client) SetTimeout(timeout time.Duration) {
+	c.timeout = timeout
+}
+
+// SetMaxAttempts changes the maximum number of attempts for TestConnection
+func (c *Client) SetMaxAttempts(attempts int) {
+	c.maxAttempts = attempts
+}
+
+// UpdateServerAddr updates the server address and resets the connection
+func (c *Client) UpdateServerAddr(serverAddr string) {
+	c.connLock.Lock()
+	defer c.connLock.Unlock()
+
+	// Close existing connection if any
+	if c.conn != nil {
+		c.conn.Close()
+		c.conn = nil
+	}
+
+	c.serverAddr = serverAddr
+}
+
+// Close cleans up client resources
+func (c *Client) Close() {
+	c.StopMonitor()
+
+	c.connLock.Lock()
+	defer c.connLock.Unlock()
+
+	if c.conn != nil {
+		c.conn.Close()
+		c.conn = nil
+	}
+}
+
+// ensureConnection makes sure we have an active UDP connection
+func (c *Client) ensureConnection() error {
+	c.connLock.Lock()
+	defer c.connLock.Unlock()
+
+	if c.conn != nil {
+		return nil
+	}
+
+	var err error
+	if c.dialer != nil {
+		c.conn, err = c.dialer("udp", c.serverAddr)
+	} else {
+		// Fallback to standard net.Dial
+		c.conn, err = net.Dial("udp", c.serverAddr)
+	}
+
+	if err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// TestConnection checks if the connection to the server is working
+// Returns true if connected, false otherwise
+func (c *Client) TestConnection(ctx context.Context) (bool, time.Duration) {
+	if err := c.ensureConnection(); err != nil {
+		logger.Warn("Failed to ensure connection: %v", err)
+		return false, 0
+	}
+
+	// Prepare packet buffer
+	packet := make([]byte, packetSize)
+	binary.BigEndian.PutUint32(packet[0:4], magicHeader)
+	packet[4] = packetTypeRequest
+
+	// Send multiple attempts as specified
+	for attempt := 0; attempt < c.maxAttempts; attempt++ {
+		select {
+		case <-ctx.Done():
+			return false, 0
+		default:
+			// Add current timestamp to packet
+			timestamp := time.Now().UnixNano()
+			binary.BigEndian.PutUint64(packet[5:13], uint64(timestamp))
+
+			// Lock the connection for the entire send/receive operation
+			c.connLock.Lock()
+
+			// Check if connection is still valid after acquiring lock
+			if c.conn == nil {
+				c.connLock.Unlock()
+				return false, 0
+			}
+
+			// logger.Debug("Attempting to send monitor packet to %s", c.serverAddr)
+			_, err := c.conn.Write(packet)
+			if err != nil {
+				c.connLock.Unlock()
+				logger.Info("Error sending packet: %v", err)
+				continue
+			}
+			// logger.Debug("Successfully sent monitor packet")
+
+			// Set read deadline
+			c.conn.SetReadDeadline(time.Now().Add(c.timeout))
+
+			// Wait for response
+			responseBuffer := make([]byte, packetSize)
+			n, err := c.conn.Read(responseBuffer)
+			c.connLock.Unlock()
+
+			if err != nil {
+				if netErr, ok := err.(net.Error); ok && netErr.Timeout() {
+					// Timeout, try next attempt
+					time.Sleep(100 * time.Millisecond) // Brief pause between attempts
+					continue
+				}
+				logger.Error("Error reading response: %v", err)
+				continue
+			}
+
+			if n != packetSize {
+				continue // Malformed packet
+			}
+
+			// Verify response
+			magic := binary.BigEndian.Uint32(responseBuffer[0:4])
+			packetType := responseBuffer[4]
+			if magic != magicHeader || packetType != packetTypeResponse {
+				continue // Not our response
+			}
+
+			// Extract the original timestamp and calculate RTT
+			sentTimestamp := int64(binary.BigEndian.Uint64(responseBuffer[5:13]))
+			rtt := time.Duration(time.Now().UnixNano() - sentTimestamp)
+
+			return true, rtt
+		}
+	}
+
+	return false, 0
+}
+
+// TestConnectionWithTimeout tries to test connection with a timeout
+// Returns true if connected, false otherwise
+func (c *Client) TestConnectionWithTimeout(timeout time.Duration) (bool, time.Duration) {
+	ctx, cancel := context.WithTimeout(context.Background(), timeout)
+	defer cancel()
+	return c.TestConnection(ctx)
+}
+
+// MonitorCallback is the function type for connection status change callbacks
+type MonitorCallback func(status ConnectionStatus)
+
+// StartMonitor begins monitoring the connection and calls the callback
+// when the connection status changes
+func (c *Client) StartMonitor(callback MonitorCallback) error {
+	c.monitorLock.Lock()
+	defer c.monitorLock.Unlock()
+
+	if c.monitorRunning {
+		logger.Info("Monitor already running")
+		return nil // Already running
+	}
+
+	if err := c.ensureConnection(); err != nil {
+		return err
+	}
+
+	c.monitorRunning = true
+	c.shutdownCh = make(chan struct{})
+
+	go func() {
+		var lastConnected bool
+		firstRun := true
+
+		ticker := time.NewTicker(c.packetInterval)
+		defer ticker.Stop()
+
+		for {
+			select {
+			case <-c.shutdownCh:
+				return
+			case <-ticker.C:
+				ctx, cancel := context.WithTimeout(context.Background(), c.timeout)
+				connected, rtt := c.TestConnection(ctx)
+				cancel()
+
+				// Callback if status changed or it's the first check
+				if connected != lastConnected || firstRun {
+					callback(ConnectionStatus{
+						Connected: connected,
+						RTT:       rtt,
+					})
+					lastConnected = connected
+					firstRun = false
+				}
+			}
+		}
+	}()
+
+	return nil
+}
+
+// StopMonitor stops the connection monitoring
+func (c *Client) StopMonitor() {
+	c.monitorLock.Lock()
+	defer c.monitorLock.Unlock()
+
+	if !c.monitorRunning {
+		return
+	}
+
+	close(c.shutdownCh)
+	c.monitorRunning = false
+}

peers/peer.go

@@ -0,0 +1,142 @@
+package peers
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/util"
+	"golang.zx2c4.com/wireguard/device"
+	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+)
+
+// ConfigurePeer sets up or updates a peer within the WireGuard device
+func ConfigurePeer(dev *device.Device, siteConfig SiteConfig, privateKey wgtypes.Key, relay bool) error {
+	var endpoint string
+	if relay && siteConfig.RelayEndpoint != "" {
+		endpoint = formatEndpoint(siteConfig.RelayEndpoint)
+	} else {
+		endpoint = formatEndpoint(siteConfig.Endpoint)
+	}
+	siteHost, err := util.ResolveDomain(endpoint)
+	if err != nil {
+		return fmt.Errorf("failed to resolve endpoint for site %d: %v", siteConfig.SiteId, err)
+	}
+
+	// Split off the CIDR of the server IP which is just a string and add /32 for the allowed IP
+	allowedIp := strings.Split(siteConfig.ServerIP, "/")
+	if len(allowedIp) > 1 {
+		allowedIp[1] = "32"
+	} else {
+		allowedIp = append(allowedIp, "32")
+	}
+	allowedIpStr := strings.Join(allowedIp, "/")
+
+	// Collect all allowed IPs in a slice
+	var allowedIPs []string
+	allowedIPs = append(allowedIPs, allowedIpStr)
+
+	// Use AllowedIps if available, otherwise fall back to RemoteSubnets for backwards compatibility
+	subnetsToAdd := siteConfig.AllowedIps
+
+	// If we have anything to add, process them
+	if len(subnetsToAdd) > 0 {
+		// Add each subnet
+		for _, subnet := range subnetsToAdd {
+			subnet = strings.TrimSpace(subnet)
+			if subnet != "" {
+				allowedIPs = append(allowedIPs, subnet)
+			}
+		}
+	}
+
+	// Construct WireGuard config for this peer
+	var configBuilder strings.Builder
+	configBuilder.WriteString(fmt.Sprintf("private_key=%s\n", util.FixKey(privateKey.String())))
+	configBuilder.WriteString(fmt.Sprintf("public_key=%s\n", util.FixKey(siteConfig.PublicKey)))
+
+	// Add each allowed IP separately
+	for _, allowedIP := range allowedIPs {
+		configBuilder.WriteString(fmt.Sprintf("allowed_ip=%s\n", allowedIP))
+	}
+
+	configBuilder.WriteString(fmt.Sprintf("endpoint=%s\n", siteHost))
+	configBuilder.WriteString("persistent_keepalive_interval=5\n")
+
+	config := configBuilder.String()
+	logger.Debug("Configuring peer with config: %s", config)
+
+	err = dev.IpcSet(config)
+	if err != nil {
+		return fmt.Errorf("failed to configure WireGuard peer: %v", err)
+	}
+
+	return nil
+}
+
+// RemovePeer removes a peer from the WireGuard device
+func RemovePeer(dev *device.Device, siteId int, publicKey string) error {
+	// Construct WireGuard config to remove the peer
+	var configBuilder strings.Builder
+	configBuilder.WriteString(fmt.Sprintf("public_key=%s\n", util.FixKey(publicKey)))
+	configBuilder.WriteString("remove=true\n")
+
+	config := configBuilder.String()
+	logger.Debug("Removing peer with config: %s", config)
+
+	err := dev.IpcSet(config)
+	if err != nil {
+		return fmt.Errorf("failed to remove WireGuard peer: %v", err)
+	}
+
+	return nil
+}
+
+// AddAllowedIP adds a single allowed IP to an existing peer without reconfiguring the entire peer
+func AddAllowedIP(dev *device.Device, publicKey string, allowedIP string) error {
+	var configBuilder strings.Builder
+	configBuilder.WriteString(fmt.Sprintf("public_key=%s\n", util.FixKey(publicKey)))
+	configBuilder.WriteString("update_only=true\n")
+	configBuilder.WriteString(fmt.Sprintf("allowed_ip=%s\n", allowedIP))
+
+	config := configBuilder.String()
+	logger.Debug("Adding allowed IP to peer with config: %s", config)
+
+	err := dev.IpcSet(config)
+	if err != nil {
+		return fmt.Errorf("failed to add allowed IP to WireGuard peer: %v", err)
+	}
+
+	return nil
+}
+
+// RemoveAllowedIP removes a single allowed IP from an existing peer by replacing the allowed IPs list
+// This requires providing all the allowed IPs that should remain after removal
+func RemoveAllowedIP(dev *device.Device, publicKey string, remainingAllowedIPs []string) error {
+	var configBuilder strings.Builder
+	configBuilder.WriteString(fmt.Sprintf("public_key=%s\n", util.FixKey(publicKey)))
+	configBuilder.WriteString("update_only=true\n")
+	configBuilder.WriteString("replace_allowed_ips=true\n")
+
+	// Add each remaining allowed IP
+	for _, allowedIP := range remainingAllowedIPs {
+		configBuilder.WriteString(fmt.Sprintf("allowed_ip=%s\n", allowedIP))
+	}
+
+	config := configBuilder.String()
+	logger.Debug("Removing allowed IP from peer with config: %s", config)
+
+	err := dev.IpcSet(config)
+	if err != nil {
+		return fmt.Errorf("failed to remove allowed IP from WireGuard peer: %v", err)
+	}
+
+	return nil
+}
+
+func formatEndpoint(endpoint string) string {
+	if strings.Contains(endpoint, ":") {
+		return endpoint
+	}
+	return endpoint + ":51820"
+}

peers/types.go

@@ -0,0 +1,64 @@
+package peers
+
+// PeerAction represents a request to add, update, or remove a peer
+type PeerAction struct {
+	Action   string     `json:"action"`   // "add", "update", or "remove"
+	SiteInfo SiteConfig `json:"siteInfo"` // Site configuration information
+}
+
+// UpdatePeerData represents the data needed to update a peer
+type SiteConfig struct {
+	SiteId        int      `json:"siteId"`
+	Name          string   `json:"name,omitempty"`
+	Endpoint      string   `json:"endpoint,omitempty"`
+	RelayEndpoint string   `json:"relayEndpoint,omitempty"`
+	PublicKey     string   `json:"publicKey,omitempty"`
+	ServerIP      string   `json:"serverIP,omitempty"`
+	ServerPort    uint16   `json:"serverPort,omitempty"`
+	RemoteSubnets []string `json:"remoteSubnets,omitempty"` // optional, array of subnets that this site can access
+	AllowedIps    []string `json:"allowedIps,omitempty"`    // optional, array of allowed IPs for the peer
+	Aliases       []Alias  `json:"aliases,omitempty"`       // optional, array of alias configurations
+}
+
+type Alias struct {
+	Alias        string `json:"alias"`        // the alias name
+	AliasAddress string `json:"aliasAddress"` // the alias IP address
+}
+
+// RemovePeer represents the data needed to remove a peer
+type PeerRemove struct {
+	SiteId int `json:"siteId"`
+}
+
+type RelayPeerData struct {
+	SiteId        int    `json:"siteId"`
+	RelayEndpoint string `json:"relayEndpoint"`
+	RelayPort     uint16 `json:"relayPort"`
+}
+
+type UnRelayPeerData struct {
+	SiteId   int    `json:"siteId"`
+	Endpoint string `json:"endpoint"`
+}
+
+// PeerAdd represents the data needed to add remote subnets to a peer
+type PeerAdd struct {
+	SiteId        int      `json:"siteId"`
+	RemoteSubnets []string `json:"remoteSubnets"`     // subnets to add
+	Aliases       []Alias  `json:"aliases,omitempty"` // aliases to add
+}
+
+// RemovePeerData represents the data needed to remove remote subnets from a peer
+type RemovePeerData struct {
+	SiteId        int      `json:"siteId"`
+	RemoteSubnets []string `json:"remoteSubnets"`     // subnets to remove
+	Aliases       []Alias  `json:"aliases,omitempty"` // aliases to remove
+}
+
+type UpdatePeerData struct {
+	SiteId           int      `json:"siteId"`
+	OldRemoteSubnets []string `json:"oldRemoteSubnets"`     // old list of remote subnets
+	NewRemoteSubnets []string `json:"newRemoteSubnets"`     // new list of remote subnets
+	OldAliases       []Alias  `json:"oldAliases,omitempty"` // old list of aliases
+	NewAliases       []Alias  `json:"newAliases,omitempty"` // new list of aliases
+}

proxy/manager.go

@@ -1,334 +0,0 @@
-package proxy
-
-import (
-	"fmt"
-	"io"
-	"net"
-	"strings"
-	"sync"
-	"time"
-
-	"github.com/fosrl/newt/logger"
-
-	"golang.zx2c4.com/wireguard/tun/netstack"
-)
-
-func NewProxyManager(tnet *netstack.Net) *ProxyManager {
-	return &ProxyManager{
-		tnet: tnet,
-	}
-}
-
-func (pm *ProxyManager) AddTarget(protocol, listen string, port int, target string) {
-	pm.Lock()
-	defer pm.Unlock()
-
-	logger.Info("Adding target: %s://%s:%d -> %s", protocol, listen, port, target)
-
-	newTarget := ProxyTarget{
-		Protocol: protocol,
-		Listen:   listen,
-		Port:     port,
-		Target:   target,
-		cancel:   make(chan struct{}),
-		done:     make(chan struct{}),
-	}
-
-	pm.targets = append(pm.targets, newTarget)
-}
-
-func (pm *ProxyManager) RemoveTarget(protocol, listen string, port int) error {
-	pm.Lock()
-	defer pm.Unlock()
-
-	protocol = strings.ToLower(protocol)
-	if protocol != "tcp" && protocol != "udp" {
-		return fmt.Errorf("unsupported protocol: %s", protocol)
-	}
-
-	for i, target := range pm.targets {
-		if target.Listen == listen &&
-			target.Port == port &&
-			strings.ToLower(target.Protocol) == protocol {
-
-			// Signal the serving goroutine to stop
-			select {
-			case <-target.cancel:
-				// Channel is already closed, no need to close it again
-			default:
-				close(target.cancel)
-			}
-
-			// Close the appropriate listener/connection based on protocol
-			target.Lock()
-			switch protocol {
-			case "tcp":
-				if target.listener != nil {
-					select {
-					case <-target.cancel:
-						// Listener was already closed by Stop()
-					default:
-						target.listener.Close()
-					}
-				}
-			case "udp":
-				if target.udpConn != nil {
-					select {
-					case <-target.cancel:
-						// Connection was already closed by Stop()
-					default:
-						target.udpConn.Close()
-					}
-				}
-			}
-			target.Unlock()
-
-			// Wait for the target to fully stop
-			<-target.done
-
-			// Remove the target from the slice
-			pm.targets = append(pm.targets[:i], pm.targets[i+1:]...)
-			return nil
-		}
-	}
-
-	return fmt.Errorf("target not found for %s %s:%d", protocol, listen, port)
-}
-
-func (pm *ProxyManager) Start() error {
-	pm.RLock()
-	defer pm.RUnlock()
-
-	for i := range pm.targets {
-		target := &pm.targets[i]
-
-		target.Lock()
-		// If target is already running, skip it
-		if target.listener != nil || target.udpConn != nil {
-			target.Unlock()
-			continue
-		}
-
-		// Mark the target as starting by creating a nil listener/connection
-		// This prevents other goroutines from trying to start it
-		if strings.ToLower(target.Protocol) == "tcp" {
-			target.listener = nil
-		} else {
-			target.udpConn = nil
-		}
-		target.Unlock()
-
-		switch strings.ToLower(target.Protocol) {
-		case "tcp":
-			go pm.serveTCP(target)
-		case "udp":
-			go pm.serveUDP(target)
-		default:
-			return fmt.Errorf("unsupported protocol: %s", target.Protocol)
-		}
-	}
-	return nil
-}
-
-func (pm *ProxyManager) Stop() error {
-	pm.Lock()
-	defer pm.Unlock()
-
-	var wg sync.WaitGroup
-	for i := range pm.targets {
-		target := &pm.targets[i]
-		wg.Add(1)
-		go func(t *ProxyTarget) {
-			defer wg.Done()
-			close(t.cancel)
-			t.Lock()
-			if t.listener != nil {
-				t.listener.Close()
-			}
-			if t.udpConn != nil {
-				t.udpConn.Close()
-			}
-			t.Unlock()
-			// Wait for the target to fully stop
-			<-t.done
-		}(target)
-	}
-	wg.Wait()
-	return nil
-}
-
-func (pm *ProxyManager) serveTCP(target *ProxyTarget) {
-	defer close(target.done) // Signal that this target is fully stopped
-
-	listener, err := pm.tnet.ListenTCP(&net.TCPAddr{
-		IP:   net.ParseIP(target.Listen),
-		Port: target.Port,
-	})
-	if err != nil {
-		logger.Info("Failed to start TCP listener for %s:%d: %v", target.Listen, target.Port, err)
-		return
-	}
-
-	target.Lock()
-	target.listener = listener
-	target.Unlock()
-
-	defer listener.Close()
-	logger.Info("TCP proxy listening on %s", listener.Addr())
-
-	var activeConns sync.WaitGroup
-	acceptDone := make(chan struct{})
-
-	// Goroutine to handle shutdown signal
-	go func() {
-		<-target.cancel
-		close(acceptDone)
-		listener.Close()
-	}()
-
-	for {
-		conn, err := listener.Accept()
-		if err != nil {
-			select {
-			case <-target.cancel:
-				// Wait for active connections to finish
-				activeConns.Wait()
-				return
-			default:
-				logger.Info("Failed to accept TCP connection: %v", err)
-				// Don't return here, try to accept new connections
-				time.Sleep(time.Second)
-				continue
-			}
-		}
-
-		activeConns.Add(1)
-		go func() {
-			defer activeConns.Done()
-			pm.handleTCPConnection(conn, target.Target, acceptDone)
-		}()
-	}
-}
-
-func (pm *ProxyManager) handleTCPConnection(clientConn net.Conn, target string, done chan struct{}) {
-	defer clientConn.Close()
-
-	serverConn, err := net.Dial("tcp", target)
-	if err != nil {
-		logger.Info("Failed to connect to target %s: %v", target, err)
-		return
-	}
-	defer serverConn.Close()
-
-	var wg sync.WaitGroup
-	wg.Add(2)
-
-	// Client -> Server
-	go func() {
-		defer wg.Done()
-		select {
-		case <-done:
-			return
-		default:
-			io.Copy(serverConn, clientConn)
-		}
-	}()
-
-	// Server -> Client
-	go func() {
-		defer wg.Done()
-		select {
-		case <-done:
-			return
-		default:
-			io.Copy(clientConn, serverConn)
-		}
-	}()
-
-	wg.Wait()
-}
-
-func (pm *ProxyManager) serveUDP(target *ProxyTarget) {
-	defer close(target.done) // Signal that this target is fully stopped
-
-	addr := &net.UDPAddr{
-		IP:   net.ParseIP(target.Listen),
-		Port: target.Port,
-	}
-
-	conn, err := pm.tnet.ListenUDP(addr)
-	if err != nil {
-		logger.Info("Failed to start UDP listener for %s:%d: %v", target.Listen, target.Port, err)
-		return
-	}
-
-	target.Lock()
-	target.udpConn = conn
-	target.Unlock()
-
-	defer conn.Close()
-	logger.Info("UDP proxy listening on %s", conn.LocalAddr())
-
-	buffer := make([]byte, 65535)
-	var activeConns sync.WaitGroup
-
-	for {
-		select {
-		case <-target.cancel:
-			activeConns.Wait() // Wait for all active UDP handlers to complete
-			return
-		default:
-			n, remoteAddr, err := conn.ReadFrom(buffer)
-			if err != nil {
-				select {
-				case <-target.cancel:
-					activeConns.Wait()
-					return
-				default:
-					logger.Info("Failed to read UDP packet: %v", err)
-					continue
-				}
-			}
-
-			targetAddr, err := net.ResolveUDPAddr("udp", target.Target)
-			if err != nil {
-				logger.Info("Failed to resolve target address %s: %v", target.Target, err)
-				continue
-			}
-
-			activeConns.Add(1)
-			go func(data []byte, remote net.Addr) {
-				defer activeConns.Done()
-				targetConn, err := net.DialUDP("udp", nil, targetAddr)
-				if err != nil {
-					logger.Info("Failed to connect to target %s: %v", target.Target, err)
-					return
-				}
-				defer targetConn.Close()
-
-				select {
-				case <-target.cancel:
-					return
-				default:
-					_, err = targetConn.Write(data)
-					if err != nil {
-						logger.Info("Failed to write to target: %v", err)
-						return
-					}
-
-					response := make([]byte, 65535)
-					n, err := targetConn.Read(response)
-					if err != nil {
-						logger.Info("Failed to read response from target: %v", err)
-						return
-					}
-
-					_, err = conn.WriteTo(response[:n], remote)
-					if err != nil {
-						logger.Info("Failed to write response to client: %v", err)
-					}
-				}
-			}(buffer[:n], remoteAddr)
-		}
-	}
-}

proxy/types.go

@@ -1,28 +0,0 @@
-package proxy
-
-import (
-	"log"
-	"net"
-	"sync"
-
-	"golang.zx2c4.com/wireguard/tun/netstack"
-)
-
-type ProxyTarget struct {
-	Protocol   string
-	Listen     string
-	Port       int
-	Target     string
-	cancel     chan struct{}  // Channel to signal shutdown
-	done       chan struct{}  // Channel to signal completion
-	listener   net.Listener   // For TCP
-	udpConn    net.PacketConn // For UDP
-	sync.Mutex                // Protect access to connection
-}
-
-type ProxyManager struct {
-	targets      []ProxyTarget
-	tnet         *netstack.Net
-	log          *log.Logger
-	sync.RWMutex // Protect access to targets slice
-}

service_unix.go

@@ -0,0 +1,54 @@
+//go:build !windows
+
+package main
+
+import (
+	"fmt"
+)
+
+// Service management functions are not available on non-Windows platforms
+func installService() error {
+	return fmt.Errorf("service management is only available on Windows")
+}
+
+func removeService() error {
+	return fmt.Errorf("service management is only available on Windows")
+}
+
+func startService(args []string) error {
+	_ = args // unused on Unix platforms
+	return fmt.Errorf("service management is only available on Windows")
+}
+
+func stopService() error {
+	return fmt.Errorf("service management is only available on Windows")
+}
+
+func getServiceStatus() (string, error) {
+	return "", fmt.Errorf("service management is only available on Windows")
+}
+
+func debugService(args []string) error {
+	_ = args // unused on Unix platforms
+	return fmt.Errorf("debug service is only available on Windows")
+}
+
+func isWindowsService() bool {
+	return false
+}
+
+func runService(name string, isDebug bool, args []string) {
+	// No-op on non-Windows platforms
+}
+
+func setupWindowsEventLog() {
+	// No-op on non-Windows platforms
+}
+
+func watchLogFile(end bool) error {
+	return fmt.Errorf("watching log file is only available on Windows")
+}
+
+func showServiceConfig() {
+	fmt.Println("Service configuration is only available on Windows")
+}

service_windows.go

@@ -0,0 +1,657 @@
+//go:build windows
+
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"log"
+	"os"
+	"os/signal"
+	"path/filepath"
+	"strings"
+	"syscall"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/sys/windows/svc"
+	"golang.org/x/sys/windows/svc/debug"
+	"golang.org/x/sys/windows/svc/eventlog"
+	"golang.org/x/sys/windows/svc/mgr"
+)
+
+const (
+	serviceName        = "OlmWireguardService"
+	serviceDisplayName = "Olm WireGuard VPN Service"
+	serviceDescription = "Olm WireGuard VPN client service for secure network connectivity"
+)
+
+// Global variable to store service arguments
+var serviceArgs []string
+
+// getServiceArgsPath returns the path where service arguments are stored
+func getServiceArgsPath() string {
+	logDir := filepath.Join(os.Getenv("PROGRAMDATA"), "olm")
+	return filepath.Join(logDir, "service_args.json")
+}
+
+// saveServiceArgs saves the service arguments to a file
+func saveServiceArgs(args []string) error {
+	logDir := filepath.Join(os.Getenv("PROGRAMDATA"), "olm")
+	err := os.MkdirAll(logDir, 0755)
+	if err != nil {
+		return fmt.Errorf("failed to create config directory: %v", err)
+	}
+
+	argsPath := getServiceArgsPath()
+	data, err := json.Marshal(args)
+	if err != nil {
+		return fmt.Errorf("failed to marshal service args: %v", err)
+	}
+
+	err = os.WriteFile(argsPath, data, 0644)
+	if err != nil {
+		return fmt.Errorf("failed to write service args: %v", err)
+	}
+
+	return nil
+}
+
+// loadServiceArgs loads the service arguments from a file
+func loadServiceArgs() ([]string, error) {
+	argsPath := getServiceArgsPath()
+	data, err := os.ReadFile(argsPath)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return []string{}, nil // Return empty args if file doesn't exist
+		}
+		return nil, fmt.Errorf("failed to read service args: %v", err)
+	}
+
+	var args []string
+	err = json.Unmarshal(data, &args)
+	if err != nil {
+		return nil, fmt.Errorf("failed to unmarshal service args: %v", err)
+	}
+
+	return args, nil
+}
+
+type olmService struct {
+	elog debug.Log
+	ctx  context.Context
+	stop context.CancelFunc
+	args []string
+}
+
+func (s *olmService) Execute(args []string, r <-chan svc.ChangeRequest, changes chan<- svc.Status) (bool, uint32) {
+	const cmdsAccepted = svc.AcceptStop | svc.AcceptShutdown
+	changes <- svc.Status{State: svc.StartPending}
+
+	s.elog.Info(1, fmt.Sprintf("Service Execute called with args: %v", args))
+
+	// Load saved service arguments
+	savedArgs, err := loadServiceArgs()
+	if err != nil {
+		s.elog.Error(1, fmt.Sprintf("Failed to load service args: %v", err))
+		// Continue with empty args if loading fails
+		savedArgs = []string{}
+	}
+	s.elog.Info(1, fmt.Sprintf("Loaded saved service args: %v", savedArgs))
+
+	// Combine service start args with saved args, giving priority to service start args
+	// Note: When the service is started via SCM, args[0] is the service name
+	// When started via s.Start(args...), the args passed are exactly what we provide
+	finalArgs := []string{}
+
+	// Check if we have args passed directly to Execute (from s.Start())
+	if len(args) > 0 {
+		// The first arg from SCM is the service name, but when we call s.Start(args...),
+		// the args we pass become args[1:] in Execute. However, if started by SCM without
+		// args, args[0] will be the service name.
+		// We need to check if args[0] looks like the service name or a flag
+		if len(args) == 1 && args[0] == serviceName {
+			// Only service name, no actual args
+			s.elog.Info(1, "Only service name in args, checking saved args")
+		} else if len(args) > 1 && args[0] == serviceName {
+			// Service name followed by actual args
+			finalArgs = append(finalArgs, args[1:]...)
+			s.elog.Info(1, fmt.Sprintf("Using service start parameters (after service name): %v", finalArgs))
+		} else {
+			// Args don't start with service name, use them all
+			// This happens when args are passed via s.Start(args...)
+			finalArgs = append(finalArgs, args...)
+			s.elog.Info(1, fmt.Sprintf("Using service start parameters (direct): %v", finalArgs))
+		}
+	}
+
+	// If no service start parameters, use saved args
+	if len(finalArgs) == 0 && len(savedArgs) > 0 {
+		finalArgs = savedArgs
+		s.elog.Info(1, fmt.Sprintf("Using saved service args: %v", finalArgs))
+	}
+
+	s.elog.Info(1, fmt.Sprintf("Final args to use: %v", finalArgs))
+	s.args = finalArgs
+
+	// Start the main olm functionality
+	olmDone := make(chan struct{})
+	go func() {
+		s.runOlm()
+		close(olmDone)
+	}()
+
+	changes <- svc.Status{State: svc.Running, Accepts: cmdsAccepted}
+	s.elog.Info(1, "Service status set to Running")
+
+	for {
+		select {
+		case c := <-r:
+			switch c.Cmd {
+			case svc.Interrogate:
+				changes <- c.CurrentStatus
+			case svc.Stop, svc.Shutdown:
+				s.elog.Info(1, "Service stopping")
+				changes <- svc.Status{State: svc.StopPending}
+				if s.stop != nil {
+					s.stop()
+				}
+				// Wait for main logic to finish or timeout
+				select {
+				case <-olmDone:
+					s.elog.Info(1, "Main logic finished gracefully")
+				case <-time.After(10 * time.Second):
+					s.elog.Info(1, "Timeout waiting for main logic to finish")
+				}
+				return false, 0
+			default:
+				s.elog.Error(1, fmt.Sprintf("Unexpected control request #%d", c))
+			}
+		case <-olmDone:
+			s.elog.Info(1, "Main olm logic completed, stopping service")
+			changes <- svc.Status{State: svc.StopPending}
+			return false, 0
+		}
+	}
+}
+
+func (s *olmService) runOlm() {
+	// Create a context that can be cancelled when the service stops
+	s.ctx, s.stop = context.WithCancel(context.Background())
+
+	// Create a separate context for programmatic shutdown (e.g., via API exit)
+	ctx, cancel := context.WithCancel(context.Background())
+
+	// Setup logging for service mode
+	s.elog.Info(1, "Starting Olm main logic")
+
+	// Run the main olm logic and wait for it to complete
+	done := make(chan struct{})
+	go func() {
+		defer func() {
+			if r := recover(); r != nil {
+				s.elog.Error(1, fmt.Sprintf("Olm panic: %v", r))
+			}
+			close(done)
+		}()
+
+		// Call the main olm function with stored arguments
+		// Use s.ctx as the signal context since the service manages shutdown
+		runOlmMainWithArgs(ctx, cancel, s.ctx, s.args)
+	}()
+
+	// Wait for either context cancellation or main logic completion
+	select {
+	case <-s.ctx.Done():
+		s.elog.Info(1, "Olm service context cancelled")
+	case <-done:
+		s.elog.Info(1, "Olm main logic completed")
+	}
+}
+
+func runService(name string, isDebug bool, args []string) {
+	var err error
+	var elog debug.Log
+
+	if isDebug {
+		elog = debug.New(name)
+		fmt.Printf("Starting %s service in debug mode\n", name)
+	} else {
+		elog, err = eventlog.Open(name)
+		if err != nil {
+			fmt.Printf("Failed to open event log: %v\n", err)
+			return
+		}
+	}
+	defer elog.Close()
+
+	elog.Info(1, fmt.Sprintf("Starting %s service", name))
+	run := svc.Run
+	if isDebug {
+		run = debug.Run
+	}
+
+	service := &olmService{elog: elog, args: args}
+	err = run(name, service)
+	if err != nil {
+		elog.Error(1, fmt.Sprintf("%s service failed: %v", name, err))
+		if isDebug {
+			fmt.Printf("Service failed: %v\n", err)
+		}
+		return
+	}
+	elog.Info(1, fmt.Sprintf("%s service stopped", name))
+	if isDebug {
+		fmt.Printf("%s service stopped\n", name)
+	}
+}
+
+func installService() error {
+	exepath, err := os.Executable()
+	if err != nil {
+		return fmt.Errorf("failed to get executable path: %v", err)
+	}
+
+	m, err := mgr.Connect()
+	if err != nil {
+		return fmt.Errorf("failed to connect to service manager: %v", err)
+	}
+	defer m.Disconnect()
+
+	s, err := m.OpenService(serviceName)
+	if err == nil {
+		s.Close()
+		return fmt.Errorf("service %s already exists", serviceName)
+	}
+
+	config := mgr.Config{
+		ServiceType:    0x10, // SERVICE_WIN32_OWN_PROCESS
+		StartType:      mgr.StartManual,
+		ErrorControl:   mgr.ErrorNormal,
+		DisplayName:    serviceDisplayName,
+		Description:    serviceDescription,
+		BinaryPathName: exepath,
+	}
+
+	s, err = m.CreateService(serviceName, exepath, config)
+	if err != nil {
+		return fmt.Errorf("failed to create service: %v", err)
+	}
+	defer s.Close()
+
+	err = eventlog.InstallAsEventCreate(serviceName, eventlog.Error|eventlog.Warning|eventlog.Info)
+	if err != nil {
+		s.Delete()
+		return fmt.Errorf("failed to install event log: %v", err)
+	}
+
+	return nil
+}
+
+func removeService() error {
+	m, err := mgr.Connect()
+	if err != nil {
+		return fmt.Errorf("failed to connect to service manager: %v", err)
+	}
+	defer m.Disconnect()
+
+	s, err := m.OpenService(serviceName)
+	if err != nil {
+		return fmt.Errorf("service %s is not installed", serviceName)
+	}
+	defer s.Close()
+
+	// Stop the service if it's running
+	status, err := s.Query()
+	if err != nil {
+		return fmt.Errorf("failed to query service status: %v", err)
+	}
+
+	if status.State != svc.Stopped {
+		_, err = s.Control(svc.Stop)
+		if err != nil {
+			return fmt.Errorf("failed to stop service: %v", err)
+		}
+
+		// Wait for service to stop
+		timeout := time.Now().Add(30 * time.Second)
+		for status.State != svc.Stopped {
+			if timeout.Before(time.Now()) {
+				return fmt.Errorf("timeout waiting for service to stop")
+			}
+			time.Sleep(300 * time.Millisecond)
+			status, err = s.Query()
+			if err != nil {
+				return fmt.Errorf("failed to query service status: %v", err)
+			}
+		}
+	}
+
+	err = s.Delete()
+	if err != nil {
+		return fmt.Errorf("failed to delete service: %v", err)
+	}
+
+	err = eventlog.Remove(serviceName)
+	if err != nil {
+		return fmt.Errorf("failed to remove event log: %v", err)
+	}
+
+	return nil
+}
+
+func startService(args []string) error {
+	fmt.Printf("Starting service with args: %v\n", args)
+
+	// Always save the service arguments so they can be loaded on service restart
+	err := saveServiceArgs(args)
+	if err != nil {
+		fmt.Printf("Warning: failed to save service args: %v\n", err)
+		// Continue anyway, args will still be passed directly
+	} else {
+		fmt.Printf("Saved service args to: %s\n", getServiceArgsPath())
+	}
+
+	m, err := mgr.Connect()
+	if err != nil {
+		return fmt.Errorf("failed to connect to service manager: %v", err)
+	}
+	defer m.Disconnect()
+
+	s, err := m.OpenService(serviceName)
+	if err != nil {
+		return fmt.Errorf("service %s is not installed", serviceName)
+	}
+	defer s.Close()
+
+	// Pass arguments directly to the service start call
+	// Note: These args will appear in Execute() after the service name
+	err = s.Start(args...)
+	if err != nil {
+		return fmt.Errorf("failed to start service: %v", err)
+	}
+
+	return nil
+}
+
+func stopService() error {
+	m, err := mgr.Connect()
+	if err != nil {
+		return fmt.Errorf("failed to connect to service manager: %v", err)
+	}
+	defer m.Disconnect()
+
+	s, err := m.OpenService(serviceName)
+	if err != nil {
+		return fmt.Errorf("service %s is not installed", serviceName)
+	}
+	defer s.Close()
+
+	status, err := s.Control(svc.Stop)
+	if err != nil {
+		return fmt.Errorf("failed to stop service: %v", err)
+	}
+
+	timeout := time.Now().Add(30 * time.Second)
+	for status.State != svc.Stopped {
+		if timeout.Before(time.Now()) {
+			return fmt.Errorf("timeout waiting for service to stop")
+		}
+		time.Sleep(300 * time.Millisecond)
+		status, err = s.Query()
+		if err != nil {
+			return fmt.Errorf("failed to query service status: %v", err)
+		}
+	}
+
+	return nil
+}
+
+func debugService(args []string) error {
+	// Save the service arguments before starting
+	if len(args) > 0 {
+		err := saveServiceArgs(args)
+		if err != nil {
+			return fmt.Errorf("failed to save service args: %v", err)
+		}
+	}
+
+	// Start the service with the provided arguments
+	err := startService(args)
+	if err != nil {
+		return fmt.Errorf("failed to start service: %v", err)
+	}
+
+	// Watch the log file
+	return watchLogFile(true)
+}
+
+func watchLogFile(end bool) error {
+	logDir := filepath.Join(os.Getenv("PROGRAMDATA"), "olm", "logs")
+	logPath := filepath.Join(logDir, "olm.log")
+
+	// Ensure the log directory exists
+	err := os.MkdirAll(logDir, 0755)
+	if err != nil {
+		return fmt.Errorf("failed to create log directory: %v", err)
+	}
+
+	// Wait for the log file to be created if it doesn't exist
+	var file *os.File
+	for i := 0; i < 30; i++ { // Wait up to 15 seconds
+		file, err = os.Open(logPath)
+		if err == nil {
+			break
+		}
+		if i == 0 {
+			fmt.Printf("Waiting for log file to be created...\n")
+		}
+		time.Sleep(500 * time.Millisecond)
+	}
+
+	if err != nil {
+		return fmt.Errorf("failed to open log file after waiting: %v", err)
+	}
+	defer file.Close()
+
+	// Seek to the end of the file to only show new logs
+	_, err = file.Seek(0, 2)
+	if err != nil {
+		return fmt.Errorf("failed to seek to end of file: %v", err)
+	}
+
+	// Set up signal handling for graceful exit
+	sigCh := make(chan os.Signal, 1)
+	signal.Notify(sigCh, os.Interrupt, syscall.SIGTERM)
+
+	// Create a ticker to check for new content
+	ticker := time.NewTicker(500 * time.Millisecond)
+	defer ticker.Stop()
+
+	buffer := make([]byte, 4096)
+
+	for {
+		select {
+		case <-sigCh:
+			fmt.Printf("\n\nStopping log watch...\n")
+			// stop the service if needed
+			if end {
+				if err := stopService(); err != nil {
+					fmt.Printf("Failed to stop service: %v\n", err)
+				}
+			}
+			fmt.Printf("Log watch stopped.\n")
+			return nil
+		case <-ticker.C:
+			// Read new content
+			n, err := file.Read(buffer)
+			if err != nil && err != io.EOF {
+				// Try to reopen the file in case it was recreated
+				file.Close()
+				file, err = os.Open(logPath)
+				if err != nil {
+					return fmt.Errorf("error reopening log file: %v", err)
+				}
+				continue
+			}
+
+			if n > 0 {
+				// Print the new content
+				fmt.Print(string(buffer[:n]))
+			}
+		}
+	}
+}
+
+func getServiceStatus() (string, error) {
+	m, err := mgr.Connect()
+	if err != nil {
+		return "", fmt.Errorf("failed to connect to service manager: %v", err)
+	}
+	defer m.Disconnect()
+
+	s, err := m.OpenService(serviceName)
+	if err != nil {
+		return "Not Installed", nil
+	}
+	defer s.Close()
+
+	status, err := s.Query()
+	if err != nil {
+		return "", fmt.Errorf("failed to query service status: %v", err)
+	}
+
+	switch status.State {
+	case svc.Stopped:
+		return "Stopped", nil
+	case svc.StartPending:
+		return "Starting", nil
+	case svc.StopPending:
+		return "Stopping", nil
+	case svc.Running:
+		return "Running", nil
+	case svc.ContinuePending:
+		return "Continue Pending", nil
+	case svc.PausePending:
+		return "Pause Pending", nil
+	case svc.Paused:
+		return "Paused", nil
+	default:
+		return "Unknown", nil
+	}
+}
+
+// showServiceConfig displays current saved service configuration
+func showServiceConfig() {
+	configPath := getServiceArgsPath()
+	fmt.Printf("Service configuration file: %s\n", configPath)
+
+	args, err := loadServiceArgs()
+	if err != nil {
+		fmt.Printf("No saved configuration found or error loading: %v\n", err)
+		return
+	}
+
+	if len(args) == 0 {
+		fmt.Println("No saved service arguments found")
+	} else {
+		fmt.Printf("Saved service arguments: %v\n", args)
+	}
+}
+
+func isWindowsService() bool {
+	isWindowsService, err := svc.IsWindowsService()
+	return err == nil && isWindowsService
+}
+
+// rotateLogFile handles daily log rotation
+func rotateLogFile(logDir string, logFile string) error {
+	// Get current log file info
+	info, err := os.Stat(logFile)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil // No current log file to rotate
+		}
+		return fmt.Errorf("failed to stat log file: %v", err)
+	}
+
+	// Check if log file is from today
+	now := time.Now()
+	fileTime := info.ModTime()
+
+	// If the log file is from today, no rotation needed
+	if now.Year() == fileTime.Year() && now.YearDay() == fileTime.YearDay() {
+		return nil
+	}
+
+	// Create rotated filename with date
+	rotatedName := fmt.Sprintf("olm-%s.log", fileTime.Format("2006-01-02"))
+	rotatedPath := filepath.Join(logDir, rotatedName)
+
+	// Rename current log file to dated filename
+	err = os.Rename(logFile, rotatedPath)
+	if err != nil {
+		return fmt.Errorf("failed to rotate log file: %v", err)
+	}
+
+	// Clean up old log files (keep last 30 days)
+	cleanupOldLogFiles(logDir, 30)
+
+	return nil
+}
+
+// cleanupOldLogFiles removes log files older than specified days
+func cleanupOldLogFiles(logDir string, daysToKeep int) {
+	cutoff := time.Now().AddDate(0, 0, -daysToKeep)
+
+	files, err := os.ReadDir(logDir)
+	if err != nil {
+		return
+	}
+
+	for _, file := range files {
+		if !file.IsDir() && strings.HasPrefix(file.Name(), "olm-") && strings.HasSuffix(file.Name(), ".log") {
+			filePath := filepath.Join(logDir, file.Name())
+			info, err := file.Info()
+			if err != nil {
+				continue
+			}
+
+			if info.ModTime().Before(cutoff) {
+				os.Remove(filePath)
+			}
+		}
+	}
+}
+
+func setupWindowsEventLog() {
+	// Create log directory if it doesn't exist
+	logDir := filepath.Join(os.Getenv("PROGRAMDATA"), "olm", "logs")
+	err := os.MkdirAll(logDir, 0755)
+	if err != nil {
+		fmt.Printf("Failed to create log directory: %v\n", err)
+		return
+	}
+
+	logFile := filepath.Join(logDir, "olm.log")
+
+	// Rotate log file if needed
+	err = rotateLogFile(logDir, logFile)
+	if err != nil {
+		fmt.Printf("Failed to rotate log file: %v\n", err)
+		// Continue anyway to create new log file
+	}
+
+	file, err := os.OpenFile(logFile, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0666)
+	if err != nil {
+		fmt.Printf("Failed to open log file: %v\n", err)
+		return
+	}
+
+	// Set the custom logger output
+	logger.GetLogger().SetOutput(file)
+
+	log.Printf("Olm service logging initialized - log file: %s", logFile)
+}

websocket/client.go

@@ -2,38 +2,108 @@ package websocket
 
 import (
 	"bytes"
+	"crypto/tls"
+	"crypto/x509"
 	"encoding/json"
 	"fmt"
+	"io"
 	"net/http"
 	"net/url"
+	"os"
 	"strings"
 	"sync"
 	"time"
 
-	"github.com/fosrl/newt/logger"
+	"software.sslmate.com/src/go-pkcs12"
 
+	"github.com/fosrl/newt/logger"
 	"github.com/gorilla/websocket"
 )
 
-type Client struct {
-	conn        *websocket.Conn
-	config      *Config
-	baseURL     string
-	handlers    map[string]MessageHandler
-	done        chan struct{}
-	handlersMux sync.RWMutex
+// AuthError represents an authentication/authorization error (401/403)
+type AuthError struct {
+	StatusCode int
+	Message    string
+}
 
+func (e *AuthError) Error() string {
+	return fmt.Sprintf("authentication error (status %d): %s", e.StatusCode, e.Message)
+}
+
+// IsAuthError checks if an error is an authentication error
+func IsAuthError(err error) bool {
+	_, ok := err.(*AuthError)
+	return ok
+}
+
+type TokenResponse struct {
+	Data struct {
+		Token         string     `json:"token"`
+		ExitNodes     []ExitNode `json:"exitNodes"`
+		ServerVersion string     `json:"serverVersion"`
+	} `json:"data"`
+	Success bool   `json:"success"`
+	Message string `json:"message"`
+}
+
+type ExitNode struct {
+	Endpoint  string `json:"endpoint"`
+	RelayPort uint16 `json:"relayPort"`
+	PublicKey string `json:"publicKey"`
+	SiteIds   []int  `json:"siteIds"`
+}
+
+type WSMessage struct {
+	Type string      `json:"type"`
+	Data interface{} `json:"data"`
+}
+
+// this is not json anymore
+type Config struct {
+	ID            string
+	Secret        string
+	Endpoint      string
+	TlsClientCert string // legacy PKCS12 file path
+	UserToken     string // optional user token for websocket authentication
+	OrgID         string // optional organization ID for websocket authentication
+}
+
+type Client struct {
+	config            *Config
+	conn              *websocket.Conn
+	baseURL           string
+	handlers          map[string]MessageHandler
+	done              chan struct{}
+	handlersMux       sync.RWMutex
 	reconnectInterval time.Duration
 	isConnected       bool
 	reconnectMux      sync.RWMutex
-
-	onConnect func() error
+	pingInterval      time.Duration
+	pingTimeout       time.Duration
+	onConnect         func() error
+	onTokenUpdate     func(token string, exitNodes []ExitNode)
+	onAuthError       func(statusCode int, message string) // Callback for auth errors
+	writeMux          sync.Mutex
+	clientType        string // Type of client (e.g., "newt", "olm")
+	tlsConfig         TLSConfig
+	configNeedsSave   bool // Flag to track if config needs to be saved
 }
 
 type ClientOption func(*Client)
 
 type MessageHandler func(message WSMessage)
 
+// TLSConfig holds TLS configuration options
+type TLSConfig struct {
+	// New separate certificate support
+	ClientCertFile string
+	ClientKeyFile  string
+	CAFiles        []string
+
+	// Existing PKCS12 support (deprecated)
+	PKCS12File string
+}
+
 // WithBaseURL sets the base URL for the client
 func WithBaseURL(url string) ClientOption {
 	return func(c *Client) {
@@ -41,16 +111,37 @@ func WithBaseURL(url string) ClientOption {
 	}
 }
 
+// WithTLSConfig sets the TLS configuration for the client
+func WithTLSConfig(config TLSConfig) ClientOption {
+	return func(c *Client) {
+		c.tlsConfig = config
+		// For backward compatibility, also set the legacy field
+		if config.PKCS12File != "" {
+			c.config.TlsClientCert = config.PKCS12File
+		}
+	}
+}
+
 func (c *Client) OnConnect(callback func() error) {
 	c.onConnect = callback
 }
 
-// NewClient creates a new Newt client
-func NewClient(newtID, secret string, endpoint string, opts ...ClientOption) (*Client, error) {
+func (c *Client) OnTokenUpdate(callback func(token string, exitNodes []ExitNode)) {
+	c.onTokenUpdate = callback
+}
+
+func (c *Client) OnAuthError(callback func(statusCode int, message string)) {
+	c.onAuthError = callback
+}
+
+// NewClient creates a new websocket client
+func NewClient(ID, secret, userToken, orgId, endpoint string, pingInterval time.Duration, pingTimeout time.Duration, opts ...ClientOption) (*Client, error) {
 	config := &Config{
-		NewtID:   newtID,
-		Secret:   secret,
-		Endpoint: endpoint,
+		ID:        ID,
+		Secret:    secret,
+		Endpoint:  endpoint,
+		UserToken: userToken,
+		OrgID:     orgId,
 	}
 
 	client := &Client{
@@ -58,39 +149,59 @@ func NewClient(newtID, secret string, endpoint string, opts ...ClientOption) (*C
 		baseURL:           endpoint, // default value
 		handlers:          make(map[string]MessageHandler),
 		done:              make(chan struct{}),
-		reconnectInterval: 10 * time.Second,
+		reconnectInterval: 3 * time.Second,
 		isConnected:       false,
+		pingInterval:      pingInterval,
+		pingTimeout:       pingTimeout,
+		clientType:        "olm",
 	}
 
 	// Apply options before loading config
 	for _, opt := range opts {
+		if opt == nil {
+			continue
+		}
 		opt(client)
 	}
 
-	// Load existing config if available
-	if err := client.loadConfig(); err != nil {
-		return nil, fmt.Errorf("failed to load config: %w", err)
-	}
-
 	return client, nil
 }
 
+func (c *Client) GetConfig() *Config {
+	return c.config
+}
+
 // Connect establishes the WebSocket connection
 func (c *Client) Connect() error {
 	go c.connectWithRetry()
 	return nil
 }
 
-// Close closes the WebSocket connection
+// Close closes the WebSocket connection gracefully
 func (c *Client) Close() error {
-	close(c.done)
-	if c.conn != nil {
-		return c.conn.Close()
+	// Signal shutdown to all goroutines first
+	select {
+	case <-c.done:
+		// Already closed
+		return nil
+	default:
+		close(c.done)
 	}
 
-	// stop the ping monitor
+	// Set connection status to false
 	c.setConnected(false)
 
+	// Close the WebSocket connection gracefully
+	if c.conn != nil {
+		// Send close message
+		c.writeMux.Lock()
+		c.conn.WriteMessage(websocket.CloseMessage, websocket.FormatCloseMessage(websocket.CloseNormalClosure, ""))
+		c.writeMux.Unlock()
+
+		// Close the connection
+		return c.conn.Close()
+	}
+
 	return nil
 }
 
@@ -105,9 +216,80 @@ func (c *Client) SendMessage(messageType string, data interface{}) error {
 		Data: data,
 	}
 
+	logger.Debug("Sending message: %s, data: %+v", messageType, data)
+
+	c.writeMux.Lock()
+	defer c.writeMux.Unlock()
 	return c.conn.WriteJSON(msg)
 }
 
+func (c *Client) SendMessageInterval(messageType string, data interface{}, interval time.Duration) (stop func(), update func(newData interface{})) {
+	stopChan := make(chan struct{})
+	updateChan := make(chan interface{})
+	var dataMux sync.Mutex
+	currentData := data
+
+	go func() {
+		count := 0
+		maxAttempts := 10
+
+		err := c.SendMessage(messageType, currentData) // Send immediately
+		if err != nil {
+			logger.Error("Failed to send initial message: %v", err)
+		}
+		count++
+
+		ticker := time.NewTicker(interval)
+		defer ticker.Stop()
+		for {
+			select {
+			case <-ticker.C:
+				if count >= maxAttempts {
+					logger.Info("SendMessageInterval timed out after %d attempts for message type: %s", maxAttempts, messageType)
+					return
+				}
+				dataMux.Lock()
+				err = c.SendMessage(messageType, currentData)
+				dataMux.Unlock()
+				if err != nil {
+					logger.Error("Failed to send message: %v", err)
+				}
+				count++
+			case newData := <-updateChan:
+				dataMux.Lock()
+				// Merge newData into currentData if both are maps
+				if currentMap, ok := currentData.(map[string]interface{}); ok {
+					if newMap, ok := newData.(map[string]interface{}); ok {
+						// Update or add keys from newData
+						for key, value := range newMap {
+							currentMap[key] = value
+						}
+						currentData = currentMap
+					} else {
+						// If newData is not a map, replace entirely
+						currentData = newData
+					}
+				} else {
+					// If currentData is not a map, replace entirely
+					currentData = newData
+				}
+				dataMux.Unlock()
+			case <-stopChan:
+				return
+			}
+		}
+	}()
+	return func() {
+			close(stopChan)
+		}, func(newData interface{}) {
+			select {
+			case updateChan <- newData:
+			case <-stopChan:
+				// Channel is closed, ignore update
+			}
+		}
+}
+
 // RegisterHandler registers a handler for a specific message type
 func (c *Client) RegisterHandler(messageType string, handler MessageHandler) {
 	c.handlersMux.Lock()
@@ -115,131 +297,109 @@ func (c *Client) RegisterHandler(messageType string, handler MessageHandler) {
 	c.handlers[messageType] = handler
 }
 
-// readPump pumps messages from the WebSocket connection
-func (c *Client) readPump() {
-	defer c.conn.Close()
-
-	for {
-		select {
-		case <-c.done:
-			return
-		default:
-			var msg WSMessage
-			err := c.conn.ReadJSON(&msg)
-			if err != nil {
-				return
-			}
-
-			c.handlersMux.RLock()
-			if handler, ok := c.handlers[msg.Type]; ok {
-				handler(msg)
-			}
-			c.handlersMux.RUnlock()
-		}
-	}
-}
-
-func (c *Client) getToken() (string, error) {
+func (c *Client) getToken() (string, []ExitNode, error) {
 	// Parse the base URL to ensure we have the correct hostname
 	baseURL, err := url.Parse(c.baseURL)
 	if err != nil {
-		return "", fmt.Errorf("failed to parse base URL: %w", err)
+		return "", nil, fmt.Errorf("failed to parse base URL: %w", err)
 	}
 
 	// Ensure we have the base URL without trailing slashes
 	baseEndpoint := strings.TrimRight(baseURL.String(), "/")
 
-	// If we already have a token, try to use it
-	if c.config.Token != "" {
-		tokenCheckData := map[string]interface{}{
-			"newtId": c.config.NewtID,
-			"secret": c.config.Secret,
-			"token":  c.config.Token,
-		}
-		jsonData, err := json.Marshal(tokenCheckData)
+	var tlsConfig *tls.Config = nil
+
+	// Use new TLS configuration method
+	if c.tlsConfig.ClientCertFile != "" || c.tlsConfig.ClientKeyFile != "" || len(c.tlsConfig.CAFiles) > 0 || c.tlsConfig.PKCS12File != "" {
+		tlsConfig, err = c.setupTLS()
 		if err != nil {
-			return "", fmt.Errorf("failed to marshal token check data: %w", err)
-		}
-
-		// Create a new request
-		req, err := http.NewRequest(
-			"POST",
-			baseEndpoint+"/api/v1/auth/newt/get-token",
-			bytes.NewBuffer(jsonData),
-		)
-		if err != nil {
-			return "", fmt.Errorf("failed to create request: %w", err)
-		}
-
-		// Set headers
-		req.Header.Set("Content-Type", "application/json")
-		req.Header.Set("X-CSRF-Token", "x-csrf-protection")
-
-		// Make the request
-		client := &http.Client{}
-		resp, err := client.Do(req)
-		if err != nil {
-			return "", fmt.Errorf("failed to check token validity: %w", err)
-		}
-		defer resp.Body.Close()
-
-		var tokenResp TokenResponse
-		if err := json.NewDecoder(resp.Body).Decode(&tokenResp); err != nil {
-			return "", fmt.Errorf("failed to decode token check response: %w", err)
-		}
-
-		// If token is still valid, return it
-		if tokenResp.Success && tokenResp.Message == "Token session already valid" {
-			return c.config.Token, nil
+			return "", nil, fmt.Errorf("failed to setup TLS configuration: %w", err)
 		}
 	}
 
-	// Get a new token
+	// Check for environment variable to skip TLS verification
+	if os.Getenv("SKIP_TLS_VERIFY") == "true" {
+		if tlsConfig == nil {
+			tlsConfig = &tls.Config{}
+		}
+		tlsConfig.InsecureSkipVerify = true
+		logger.Debug("TLS certificate verification disabled via SKIP_TLS_VERIFY environment variable")
+	}
+
 	tokenData := map[string]interface{}{
-		"newtId": c.config.NewtID,
+		"olmId":  c.config.ID,
 		"secret": c.config.Secret,
+		"orgId":  c.config.OrgID,
 	}
 	jsonData, err := json.Marshal(tokenData)
+
 	if err != nil {
-		return "", fmt.Errorf("failed to marshal token request data: %w", err)
+		return "", nil, fmt.Errorf("failed to marshal token request data: %w", err)
 	}
 
 	// Create a new request
 	req, err := http.NewRequest(
 		"POST",
-		baseEndpoint+"/api/v1/auth/newt/get-token",
+		baseEndpoint+"/api/v1/auth/"+c.clientType+"/get-token",
 		bytes.NewBuffer(jsonData),
 	)
 	if err != nil {
-		return "", fmt.Errorf("failed to create request: %w", err)
+		return "", nil, fmt.Errorf("failed to create request: %w", err)
 	}
 
 	// Set headers
 	req.Header.Set("Content-Type", "application/json")
 	req.Header.Set("X-CSRF-Token", "x-csrf-protection")
 
+	// print out the request for debugging
+	logger.Debug("Requesting token from %s with body: %s", req.URL.String(), string(jsonData))
+
 	// Make the request
 	client := &http.Client{}
+	if tlsConfig != nil {
+		client.Transport = &http.Transport{
+			TLSClientConfig: tlsConfig,
+		}
+	}
 	resp, err := client.Do(req)
 	if err != nil {
-		return "", fmt.Errorf("failed to request new token: %w", err)
+		return "", nil, fmt.Errorf("failed to request new token: %w", err)
 	}
 	defer resp.Body.Close()
 
+	if resp.StatusCode != http.StatusOK {
+		body, _ := io.ReadAll(resp.Body)
+		logger.Error("Failed to get token with status code: %d, body: %s", resp.StatusCode, string(body))
+
+		// Return AuthError for 401/403 status codes
+		if resp.StatusCode == http.StatusUnauthorized || resp.StatusCode == http.StatusForbidden {
+			return "", nil, &AuthError{
+				StatusCode: resp.StatusCode,
+				Message:    string(body),
+			}
+		}
+
+		// For other errors (5xx, network issues, etc.), return regular error
+		return "", nil, fmt.Errorf("failed to get token with status code: %d, body: %s", resp.StatusCode, string(body))
+	}
+
 	var tokenResp TokenResponse
 	if err := json.NewDecoder(resp.Body).Decode(&tokenResp); err != nil {
-		return "", fmt.Errorf("failed to decode token response: %w", err)
+		logger.Error("Failed to decode token response.")
+		return "", nil, fmt.Errorf("failed to decode token response: %w", err)
 	}
 
 	if !tokenResp.Success {
-		return "", fmt.Errorf("failed to get token: %s", tokenResp.Message)
+		return "", nil, fmt.Errorf("failed to get token: %s", tokenResp.Message)
 	}
 
 	if tokenResp.Data.Token == "" {
-		return "", fmt.Errorf("received empty token from server")
+		return "", nil, fmt.Errorf("received empty token from server")
 	}
 
-	return tokenResp.Data.Token, nil
+	logger.Debug("Received token: %s", tokenResp.Data.Token)
+
+	return tokenResp.Data.Token, tokenResp.Data.ExitNodes, nil
 }
 
 func (c *Client) connectWithRetry() {
@@ -250,6 +410,18 @@ func (c *Client) connectWithRetry() {
 		default:
 			err := c.establishConnection()
 			if err != nil {
+				// Check if this is an auth error (401/403)
+				if authErr, ok := err.(*AuthError); ok {
+					logger.Error("Authentication failed: %v. Terminating tunnel and retrying...", authErr)
+					// Trigger auth error callback if set (this should terminate the tunnel)
+					if c.onAuthError != nil {
+						c.onAuthError(authErr.StatusCode, authErr.Message)
+					}
+					// Continue retrying after auth error
+					time.Sleep(c.reconnectInterval)
+					continue
+				}
+				// For other errors (5xx, network issues), continue retrying
 				logger.Error("Failed to connect: %v. Retrying in %v...", err, c.reconnectInterval)
 				time.Sleep(c.reconnectInterval)
 				continue
@@ -261,11 +433,15 @@ func (c *Client) connectWithRetry() {
 
 func (c *Client) establishConnection() error {
 	// Get token for authentication
-	token, err := c.getToken()
+	token, exitNodes, err := c.getToken()
 	if err != nil {
 		return fmt.Errorf("failed to get token: %w", err)
 	}
 
+	if c.onTokenUpdate != nil {
+		c.onTokenUpdate(token, exitNodes)
+	}
+
 	// Parse the base URL to determine protocol and hostname
 	baseURL, err := url.Parse(c.baseURL)
 	if err != nil {
@@ -288,10 +464,35 @@ func (c *Client) establishConnection() error {
 	// Add token to query parameters
 	q := u.Query()
 	q.Set("token", token)
+	q.Set("clientType", c.clientType)
+	if c.config.UserToken != "" {
+		q.Set("userToken", c.config.UserToken)
+	}
 	u.RawQuery = q.Encode()
 
 	// Connect to WebSocket
-	conn, _, err := websocket.DefaultDialer.Dial(u.String(), nil)
+	dialer := websocket.DefaultDialer
+
+	// Use new TLS configuration method
+	if c.tlsConfig.ClientCertFile != "" || c.tlsConfig.ClientKeyFile != "" || len(c.tlsConfig.CAFiles) > 0 || c.tlsConfig.PKCS12File != "" {
+		logger.Info("Setting up TLS configuration for WebSocket connection")
+		tlsConfig, err := c.setupTLS()
+		if err != nil {
+			return fmt.Errorf("failed to setup TLS configuration: %w", err)
+		}
+		dialer.TLSClientConfig = tlsConfig
+	}
+
+	// Check for environment variable to skip TLS verification for WebSocket connection
+	if os.Getenv("SKIP_TLS_VERIFY") == "true" {
+		if dialer.TLSClientConfig == nil {
+			dialer.TLSClientConfig = &tls.Config{}
+		}
+		dialer.TLSClientConfig.InsecureSkipVerify = true
+		logger.Debug("WebSocket TLS certificate verification disabled via SKIP_TLS_VERIFY environment variable")
+	}
+
+	conn, _, err := dialer.Dial(u.String(), nil)
 	if err != nil {
 		return fmt.Errorf("failed to connect to WebSocket: %w", err)
 	}
@@ -301,8 +502,8 @@ func (c *Client) establishConnection() error {
 
 	// Start the ping monitor
 	go c.pingMonitor()
-	// Start the read pump
-	go c.readPump()
+	// Start the read pump with disconnect detection
+	go c.readPumpWithDisconnectDetection()
 
 	if c.onConnect != nil {
 		if err := c.onConnect(); err != nil {
@@ -313,8 +514,72 @@ func (c *Client) establishConnection() error {
 	return nil
 }
 
+// setupTLS configures TLS based on the TLS configuration
+func (c *Client) setupTLS() (*tls.Config, error) {
+	tlsConfig := &tls.Config{}
+
+	// Handle new separate certificate configuration
+	if c.tlsConfig.ClientCertFile != "" && c.tlsConfig.ClientKeyFile != "" {
+		logger.Info("Loading separate certificate files for mTLS")
+		logger.Debug("Client cert: %s", c.tlsConfig.ClientCertFile)
+		logger.Debug("Client key: %s", c.tlsConfig.ClientKeyFile)
+
+		// Load client certificate and key
+		cert, err := tls.LoadX509KeyPair(c.tlsConfig.ClientCertFile, c.tlsConfig.ClientKeyFile)
+		if err != nil {
+			return nil, fmt.Errorf("failed to load client certificate pair: %w", err)
+		}
+		tlsConfig.Certificates = []tls.Certificate{cert}
+
+		// Load CA certificates for remote validation if specified
+		if len(c.tlsConfig.CAFiles) > 0 {
+			logger.Debug("Loading CA certificates: %v", c.tlsConfig.CAFiles)
+			caCertPool := x509.NewCertPool()
+			for _, caFile := range c.tlsConfig.CAFiles {
+				caCert, err := os.ReadFile(caFile)
+				if err != nil {
+					return nil, fmt.Errorf("failed to read CA file %s: %w", caFile, err)
+				}
+
+				// Try to parse as PEM first, then DER
+				if !caCertPool.AppendCertsFromPEM(caCert) {
+					// If PEM parsing failed, try DER
+					cert, err := x509.ParseCertificate(caCert)
+					if err != nil {
+						return nil, fmt.Errorf("failed to parse CA certificate from %s: %w", caFile, err)
+					}
+					caCertPool.AddCert(cert)
+				}
+			}
+			tlsConfig.RootCAs = caCertPool
+		}
+
+		return tlsConfig, nil
+	}
+
+	// Fallback to existing PKCS12 implementation for backward compatibility
+	if c.tlsConfig.PKCS12File != "" {
+		logger.Info("Loading PKCS12 certificate for mTLS (deprecated)")
+		return c.setupPKCS12TLS()
+	}
+
+	// Legacy fallback using config.TlsClientCert
+	if c.config.TlsClientCert != "" {
+		logger.Info("Loading legacy PKCS12 certificate for mTLS (deprecated)")
+		return loadClientCertificate(c.config.TlsClientCert)
+	}
+
+	return nil, nil
+}
+
+// setupPKCS12TLS loads TLS configuration from PKCS12 file
+func (c *Client) setupPKCS12TLS() (*tls.Config, error) {
+	return loadClientCertificate(c.tlsConfig.PKCS12File)
+}
+
+// pingMonitor sends pings at a short interval and triggers reconnect on failure
 func (c *Client) pingMonitor() {
-	ticker := time.NewTicker(30 * time.Second)
+	ticker := time.NewTicker(c.pingInterval)
 	defer ticker.Stop()
 
 	for {
@@ -322,11 +587,74 @@ func (c *Client) pingMonitor() {
 		case <-c.done:
 			return
 		case <-ticker.C:
-			if err := c.conn.WriteControl(websocket.PingMessage, []byte{}, time.Now().Add(10*time.Second)); err != nil {
-				logger.Error("Ping failed: %v", err)
-				c.reconnect()
+			if c.conn == nil {
 				return
 			}
+			c.writeMux.Lock()
+			err := c.conn.WriteControl(websocket.PingMessage, []byte{}, time.Now().Add(c.pingTimeout))
+			c.writeMux.Unlock()
+			if err != nil {
+				// Check if we're shutting down before logging error and reconnecting
+				select {
+				case <-c.done:
+					// Expected during shutdown
+					return
+				default:
+					logger.Error("Ping failed: %v", err)
+					c.reconnect()
+					return
+				}
+			}
+		}
+	}
+}
+
+// readPumpWithDisconnectDetection reads messages and triggers reconnect on error
+func (c *Client) readPumpWithDisconnectDetection() {
+	defer func() {
+		if c.conn != nil {
+			c.conn.Close()
+		}
+		// Only attempt reconnect if we're not shutting down
+		select {
+		case <-c.done:
+			// Shutting down, don't reconnect
+			return
+		default:
+			c.reconnect()
+		}
+	}()
+
+	for {
+		select {
+		case <-c.done:
+			return
+		default:
+			var msg WSMessage
+			err := c.conn.ReadJSON(&msg)
+			if err != nil {
+				// Check if we're shutting down before logging error
+				select {
+				case <-c.done:
+					// Expected during shutdown, don't log as error
+					logger.Debug("WebSocket connection closed during shutdown")
+					return
+				default:
+					// Unexpected error during normal operation
+					if websocket.IsUnexpectedCloseError(err, websocket.CloseGoingAway, websocket.CloseAbnormalClosure, websocket.CloseNormalClosure) {
+						logger.Error("WebSocket read error: %v", err)
+					} else {
+						logger.Debug("WebSocket connection closed: %v", err)
+					}
+					return // triggers reconnect via defer
+				}
+			}
+
+			c.handlersMux.RLock()
+			if handler, ok := c.handlers[msg.Type]; ok {
+				handler(msg)
+			}
+			c.handlersMux.RUnlock()
 		}
 	}
 }
@@ -335,9 +663,16 @@ func (c *Client) reconnect() {
 	c.setConnected(false)
 	if c.conn != nil {
 		c.conn.Close()
+		c.conn = nil
 	}
 
-	go c.connectWithRetry()
+	// Only reconnect if we're not shutting down
+	select {
+	case <-c.done:
+		return
+	default:
+		go c.connectWithRetry()
+	}
 }
 
 func (c *Client) setConnected(status bool) {
@@ -345,3 +680,42 @@ func (c *Client) setConnected(status bool) {
 	defer c.reconnectMux.Unlock()
 	c.isConnected = status
 }
+
+// LoadClientCertificate Helper method to load client certificates (PKCS12 format)
+func loadClientCertificate(p12Path string) (*tls.Config, error) {
+	logger.Info("Loading tls-client-cert %s", p12Path)
+	// Read the PKCS12 file
+	p12Data, err := os.ReadFile(p12Path)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read PKCS12 file: %w", err)
+	}
+
+	// Parse PKCS12 with empty password for non-encrypted files
+	privateKey, certificate, caCerts, err := pkcs12.DecodeChain(p12Data, "")
+	if err != nil {
+		return nil, fmt.Errorf("failed to decode PKCS12: %w", err)
+	}
+
+	// Create certificate
+	cert := tls.Certificate{
+		Certificate: [][]byte{certificate.Raw},
+		PrivateKey:  privateKey,
+	}
+
+	// Optional: Add CA certificates if present
+	rootCAs, err := x509.SystemCertPool()
+	if err != nil {
+		return nil, fmt.Errorf("failed to load system cert pool: %w", err)
+	}
+	if len(caCerts) > 0 {
+		for _, caCert := range caCerts {
+			rootCAs.AddCert(caCert)
+		}
+	}
+
+	// Create TLS configuration
+	return &tls.Config{
+		Certificates: []tls.Certificate{cert},
+		RootCAs:      rootCAs,
+	}, nil
+}

websocket/config.go

@@ -1,72 +0,0 @@
-package websocket
-
-import (
-	"encoding/json"
-	"log"
-	"os"
-	"path/filepath"
-	"runtime"
-)
-
-func getConfigPath() string {
-	var configDir string
-	switch runtime.GOOS {
-	case "darwin":
-		configDir = filepath.Join(os.Getenv("HOME"), "Library", "Application Support", "newt-client")
-	case "windows":
-		configDir = filepath.Join(os.Getenv("APPDATA"), "newt-client")
-	default: // linux and others
-		configDir = filepath.Join(os.Getenv("HOME"), ".config", "newt-client")
-	}
-
-	if err := os.MkdirAll(configDir, 0755); err != nil {
-		log.Printf("Failed to create config directory: %v", err)
-	}
-
-	return filepath.Join(configDir, "config.json")
-}
-
-func (c *Client) loadConfig() error {
-	if c.config.NewtID != "" && c.config.Secret != "" && c.config.Endpoint != "" {
-		return nil
-	}
-
-	configPath := getConfigPath()
-	data, err := os.ReadFile(configPath)
-	if err != nil {
-		if os.IsNotExist(err) {
-			return nil
-		}
-		return err
-	}
-
-	var config Config
-	if err := json.Unmarshal(data, &config); err != nil {
-		return err
-	}
-
-	if c.config.NewtID == "" {
-		c.config.NewtID = config.NewtID
-	}
-	if c.config.Token == "" {
-		c.config.Token = config.Token
-	}
-	if c.config.Secret == "" {
-		c.config.Secret = config.Secret
-	}
-	if c.config.Endpoint == "" {
-		c.config.Endpoint = config.Endpoint
-		c.baseURL = config.Endpoint
-	}
-
-	return nil
-}
-
-func (c *Client) saveConfig() error {
-	configPath := getConfigPath()
-	data, err := json.MarshalIndent(c.config, "", "  ")
-	if err != nil {
-		return err
-	}
-	return os.WriteFile(configPath, data, 0644)
-}

websocket/types.go

@@ -1,21 +0,0 @@
-package websocket
-
-type Config struct {
-	NewtID   string `json:"newtId"`
-	Secret   string `json:"secret"`
-	Token    string `json:"token"`
-	Endpoint string `json:"endpoint"`
-}
-
-type TokenResponse struct {
-	Data struct {
-		Token string `json:"token"`
-	} `json:"data"`
-	Success bool   `json:"success"`
-	Message string `json:"message"`
-}
-
-type WSMessage struct {
-	Type string      `json:"type"`
-	Data interface{} `json:"data"`
-}