fleet-and-orbit.yml

# This workflow tests orbit code changes (compiles orbit from source).
# It uses a fleet instance also built and executed from source.
#
# It tests that orbit osquery agents enroll successfully to Fleet.
name: Test Fleetctl Package, Orbit & Fleet

on:
  push:
    branches:
      - main
      - patch-*
      - prepare-*
    paths:
      - "orbit/**.go"
      - ".github/workflows/fleet-and-orbit.yml"
  pull_request:
    paths:
      - "orbit/**.go"
      - ".github/workflows/fleet-and-orbit.yml"
  workflow_dispatch: # Manual

# This allows a subsequently queued workflow run to interrupt previous runs
concurrency:
  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id}}
  cancel-in-progress: true

defaults:
  run:
    # fail-fast using bash -eo pipefail. See https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#exit-codes-and-error-action-preference
    shell: bash

env:
  OSQUERY_VERSION: 5.9.1

permissions:
  contents: read

jobs:
  gen:
    runs-on: ubuntu-latest
    outputs:
      subdomain: ${{ steps.gen.outputs.subdomain }}
      domain: ${{ steps.gen.outputs.domain }}
      address: ${{ steps.gen.outputs.address }}
      enroll_secret: ${{ steps.gen.outputs.enroll_secret }}
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - id: gen
        run: |
          UUID=$(uuidgen)
          echo "subdomain=fleet-test-$UUID" >> $GITHUB_OUTPUT
          echo "domain=fleet-test-$UUID.fleetuem.com" >> $GITHUB_OUTPUT
          echo "address=https://fleet-test-$UUID.fleetuem.com" >> $GITHUB_OUTPUT
          ENROLL=$(uuidgen)
          echo "enroll_secret=$ENROLL" >> $GITHUB_OUTPUT

  run-server:
    timeout-minutes: 60
    strategy:
      matrix:
        mysql: ["mysql:8.0.36"]
    runs-on: ubuntu-latest
    needs: gen
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Checkout Code
        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9 # v3.5.3

      - name: Install Go
        uses: actions/setup-go@93397bea11091df50f3d7e59dc26a7711a8bcfbe # v4.1.0
        with:
          go-version-file: 'go.mod'

      - name: Set up Node.js
        uses: actions/setup-node@5e21ff4d9bc1a8cf6de233a3057d20ec6b3fb69d # v3.8.1
        with:
          node-version-file: package.json

      - name: Start tunnel
        env:
          CERT_PEM: ${{ secrets.CLOUDFLARE_TUNNEL_FLEETUEM_CERT_B64 }}
        run: |
          #!/bin/bash
          # Increase maximum receive buffer size to roughly 2.5 MB.
          # Cloudflared uses quic-go. This buffer holds packets that have been received by the kernel,
          # but not yet read by the application (quic-go in this case). Once this buffer fills up, the
          # kernel will drop any new incoming packet.
          # See https://github.com/quic-go/quic-go/wiki/UDP-Receive-Buffer-Size.
          sudo sysctl -w net.core.rmem_max=2500000

          # Install cloudflared
          wget https://github.com/cloudflare/cloudflared/releases/latest/download/cloudflared-linux-amd64.deb
          sudo dpkg -i cloudflared-linux-amd64.deb
          # Add secret
          echo "$CERT_PEM" | base64 -d > cert.pem
          # Start tunnel
          cloudflared tunnel --origincert cert.pem --hostname ${{ needs.gen.outputs.subdomain }} --url http://localhost:1337 --name ${{ needs.gen.outputs.subdomain }} --logfile cloudflared.log &
          until [[ $(cloudflared tunnel --origincert cert.pem info -o json ${{ needs.gen.outputs.subdomain }} | jq '.conns[0].conns[0].is_pending_reconnect') = false ]]; do
            echo "Awaiting tunnel ready..."
            sleep 5
          done

      - name: Start Infra Dependencies
        run: FLEET_MYSQL_IMAGE=${{ matrix.mysql }} docker compose up -d mysql redis &

      - name: Install JS Dependencies
        run: make deps-js

      - name: Generate and bundle go & js code
        run: make generate

      - name: Build fleet and fleetctl
        # fleet-dev builds fleet with "race" enabled.
        run: make fleet-dev fleetctl

      - name: Run Fleet server
        env:
          FLEET_OSQUERY_HOST_IDENTIFIER: instance # use instance identifier to allow for duplicate UUIDs
          FLEET_SERVER_ADDRESS: 0.0.0.0:1337
          FLEET_SERVER_TLS: false
          FLEET_LOGGING_DEBUG: true
        run: |
          mkdir ./fleet_log
          make db-reset
          ./build/fleet serve --dev --dev_license 1>./fleet_log/stdout.log 2>./fleet_log/stderr.log &
          ./build/fleetctl config set --address http://localhost:1337 --tls-skip-verify
          until ./build/fleetctl setup --email admin@example.com --name Admin --password preview1337# --org-name Example
          do
            echo "Retrying setup in 5s..."
            sleep 5
          done
          # Wait for all of the hosts to be enrolled
          EXPECTED=3
          until [ $(./build/fleetctl get hosts --json | grep "hostname" | wc -l | tee hostcount) -ge $EXPECTED ]; do
            echo -n "Waiting for hosts to enroll: "
            cat hostcount | xargs echo -n
            echo " / $EXPECTED"
            sleep 30
          done
          ./build/fleetctl get hosts
          ./build/fleetctl get hosts --json
          echo "Success! $EXPECTED hosts enrolled."

      - name: Cleanup tunnel
        if: always()
        run: cloudflared tunnel --origincert cert.pem delete --force ${{ needs.gen.outputs.subdomain }}

      - name: Upload fleet logs
        if: always()
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: fleet-logs
          path: |
            fleet_log

      - name: Upload cloudflared logs
        if: always()
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: cloudflared.log
          path: cloudflared.log

  # Sets the enroll secret of the Fleet server.
  #
  # This job also makes sure the Fleet server is up and running.
  set-enroll-secret:
    timeout-minutes: 60
    runs-on: ubuntu-latest
    needs: gen
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Checkout Code
        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9 # v3.5.3

      - name: Install Go
        uses: actions/setup-go@93397bea11091df50f3d7e59dc26a7711a8bcfbe # v4.1.0
        with:
          go-version-file: 'go.mod'

      - name: Build Fleetctl
        run: make fleetctl

      - id: enroll
        name: Set enroll secret
        run: |
          ./build/fleetctl config set --address ${{ needs.gen.outputs.address }}
          until ./build/fleetctl login --email admin@example.com --password preview1337#
          do
            echo "Retrying in 30s..."
            sleep 30
          done
          echo '---
          apiVersion: v1
          kind: enroll_secret
          spec:
            secrets:
            - secret: ${{ needs.gen.outputs.enroll_secret }}
          ' > secrets.yml
          ./build/fleetctl apply -f secrets.yml

  # Here we generate the Fleet Desktop and osqueryd targets for
  # macOS which can only be generated from a macOS host.
  build-macos-targets:
    # Set macOS version to '12' (current equivalent to macos-latest) for
    # building the binary. This ensures compatibility with macOS version 13 and
    # later, avoiding runtime errors on systems using macOS 13 or newer.
    #
    # Note: Update this version to '13' once GitHub marks macOS 13 as stable
    # or if we revise our minimum supported macOS version.
    runs-on: macos-12
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Checkout Code
        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9 # v3.5.3

      - name: Install Go
        uses: actions/setup-go@93397bea11091df50f3d7e59dc26a7711a8bcfbe # v4.1.0
        with:
          go-version-file: 'go.mod'

      - name: Build desktop.app.tar.gz and osqueryd.app.tar.gz
        run: |
          make desktop-app-tar-gz
          make osqueryd-app-tar-gz version=$OSQUERY_VERSION out-path=.

      - name: Upload desktop.app.tar.gz and osqueryd.app.tar.gz
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: macos-pre-built-apps
          path: |
            desktop.app.tar.gz
            osqueryd.app.tar.gz

  # TODO(lucas): Currently, to simplify the workflow we do all in one job:
  # 1. Generate TUF repository (compile Orbit from source).
  # 2. Run TUF server on localhost.
  # 3. Generate packages using localhost TUF server.
  #
  # When installing the generated packages, Orbit will log "update errors"
  # because the TUF URL is set to http://localhost:8081.
  #
  # TODO(lucas): Test the generated RPM package on a CentOS docker image.
  #
  # We run this job in ubuntu because Github macOS runner doesn't have Docker
  # installed, and installing it is time consuming and unreliable.
  run-tuf-and-gen-pkgs:
    timeout-minutes: 60
    runs-on: ubuntu-latest
    needs: [gen, build-macos-targets]
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Checkout Code
        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9 # v3.5.3

      - name: Install Go
        uses: actions/setup-go@93397bea11091df50f3d7e59dc26a7711a8bcfbe # v4.1.0
        with:
          go-version-file: 'go.mod'

      - name: Download macos pre-built apps
        id: download
        uses: actions/download-artifact@fb598a63ae348fa914e94cd0ff38f362e927b741 # v2
        with:
          name: macos-pre-built-apps

      - name: Build Repository and run TUF server
        env:
          SYSTEMS: "macos windows linux linux-arm64"
          PKG_FLEET_URL: ${{ needs.gen.outputs.address }}
          PKG_TUF_URL: http://localhost:8081
          DEB_FLEET_URL: ${{ needs.gen.outputs.address }}
          DEB_TUF_URL: http://localhost:8081
          RPM_FLEET_URL: ${{ needs.gen.outputs.address }}
          RPM_TUF_URL: http://localhost:8081
          MSI_FLEET_URL: ${{ needs.gen.outputs.address }}
          MSI_TUF_URL: http://localhost:8081
          ENROLL_SECRET: ${{ needs.gen.outputs.enroll_secret }}
          MACOS_USE_PREBUILT_DESKTOP_APP_TAR_GZ: 1
          MACOS_USE_PREBUILT_OSQUERYD_APP_TAR_GZ: 1
          GENERATE_PKG: 1
          GENERATE_DEB: 1
          GENERATE_RPM: 1
          GENERATE_MSI: 1
          FLEET_DESKTOP: 1
        run: |
          ./tools/tuf/test/main.sh

      - name: Upload PKG installer
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: fleet-osquery.pkg
          path: |
            fleet-osquery.pkg

      - name: Upload DEB installer
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: fleet-osquery_42.0.0_amd64.deb
          path: |
            fleet-osquery_42.0.0_amd64.deb

      - name: Upload MSI installer
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: fleet-osquery.msi
          path: |
            fleet-osquery.msi

  orbit-macos:
    timeout-minutes: 60
    runs-on: macos-latest
    needs: [gen, run-tuf-and-gen-pkgs]
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Checkout Code
        uses: actions/checkout@c85c95e3d7251135ab7dc9ce3241c5835cc595a9 # v3.5.3

      - name: Download pkg
        id: download
        uses: actions/download-artifact@fb598a63ae348fa914e94cd0ff38f362e927b741 # v2
        with:
          name: fleet-osquery.pkg

      - name: Install pkg
        run: |
          sudo hostname orbit-macos
          sudo installer -pkg ${{ steps.download.outputs.download-path }}/fleet-osquery.pkg -target /

      - name: Wait enroll
        run: |
          # Wait until fleet server goes down.
          while curl --fail ${{ needs.gen.outputs.address }};
          do
            echo "Retrying in 10s..."
            sleep 10
          done

      - name: Run orbit shell
        run: sudo orbit shell -- --json "select * from osquery_info;" | jq -e 'if (.[0]) then true else false end'

      - name: Collect orbit logs
        if: always()
        run: |
          mkdir orbit-logs
          sudo cp /var/log/orbit/* orbit-logs/

      - name: Upload orbit logs
        if: always()
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: orbit-logs
          path: |
            orbit-logs

      - name: Uninstall pkg
        run: |
          sudo ./orbit/tools/cleanup/cleanup_macos.sh

  orbit-ubuntu:
    timeout-minutes: 60
    runs-on: ubuntu-latest
    needs: [gen, run-tuf-and-gen-pkgs]
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Download deb
        id: download
        uses: actions/download-artifact@fb598a63ae348fa914e94cd0ff38f362e927b741 # v2
        with:
          name: fleet-osquery_42.0.0_amd64.deb

      - name: Install deb
        run: |
          sudo hostname orbit-ubuntu
          sudo dpkg --install ${{ steps.download.outputs.download-path }}/fleet-osquery_42.0.0_amd64.deb

      - name: Wait enroll
        run: |
          # Wait until fleet server goes down.
          while curl --fail ${{ needs.gen.outputs.address }};
          do
            echo "Retrying in 10s..."
            sleep 10
          done

      - name: Run orbit shell
        run: sudo orbit shell -- --json "select * from osquery_info;" | jq -e 'if (.[0]) then true else false end'

      - name: Collect orbit logs
        if: always()
        run: |
          mkdir orbit-logs
          sudo journalctl -u orbit.service > orbit-logs/orbit_service.log

      - name: Upload orbit logs
        if: always()
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: orbit-logs
          path: |
            orbit-logs

      - name: Uninstall deb
        run: |
          sudo apt remove fleet-osquery -y

  orbit-windows:
    timeout-minutes: 60
    needs: [gen, run-tuf-and-gen-pkgs]
    runs-on: windows-latest
    steps:
      - name: Harden Runner
        uses: step-security/harden-runner@63c24ba6bd7ba022e95695ff85de572c04a18142 # v2.7.0
        with:
          egress-policy: audit

      - name: Download msi
        id: download
        uses: actions/download-artifact@fb598a63ae348fa914e94cd0ff38f362e927b741 # v2
        with:
          name: fleet-osquery.msi

      - name: Install msi
        shell: pwsh
        run: |
          Start-Process msiexec -ArgumentList "/i ${{ steps.download.outputs.download-path }}\fleet-osquery.msi /quiet /passive /lv log.txt" -Wait

      - name: Wait enroll
        run: |
          while curl --fail ${{ needs.gen.outputs.address }};
          do
            echo "Retrying in 10s..."
            sleep 10
          done

      - name: Run orbit shell
        shell: cmd
        run: |
          "C:\Program Files\Orbit\bin\orbit\orbit.exe" shell -- --json "select * from osquery_info;" | jq -e "if (.[0]) then true else false end"

      - name: Fleet Service Tests
        shell: pwsh
        run: |
          # Tests setup
          $serviceName = "Fleet osquery"
          $orbitMaxTimeToStartAndTeardown = 15

          # Test 1 - Check that the service starts without issues
          Stop-Service -Name $serviceName
          Start-Sleep -Seconds $orbitMaxTimeToStartAndTeardown
          Start-Service -Name $serviceName
          Get-Service -Name $serviceName | %{ if ($_.Status -ne "Running") { throw "Fleet Service test #1 failed" } }

          # Test 2 - Check that the service stops without issues
          Stop-Service -Name $serviceName
          Start-Sleep -Seconds $orbitMaxTimeToStartAndTeardown
          Get-Service -Name $serviceName | %{ if ($_.Status -ne "Stopped") { throw "Fleet Service test #2 failed" } }

          # Test 3 - Check that no orbit.exe is running after service stop (updated after graceful shutdown)
          #Start-Service -Name $serviceName
          #Start-Sleep -Seconds $orbitMaxTimeToStartAndTeardown
          #Stop-Service -Name $serviceName
          #Start-Sleep -Seconds ($orbitMaxTimeToStartAndTeardown * 10) # there is an issue with osqueryd runner intertupt that needs to be tracked down
          #Get-Process | %{ if ($_.Name -eq "orbit") { throw "Fleet Service test #3 failed" } }

          # Test 4 - Check that service starts in less than 3 secs
          #Start-Job { Start-Service -Name $args[0] } -ArgumentList $serviceName | Out-Null #async operation
          #Start-Sleep -Seconds 3
          #Get-Service -Name $serviceName | %{ if ($_.Status -ne "Running") { throw "Fleet Service test #4 failed" } }

          # Test 5 - Check that service stops in less than $orbitMaxTimeToStartAndTeardown secs
          #Start-Job { Stop-Service -Name $args[0] } -ArgumentList $serviceName  | Out-Null #async operation
          #Start-Sleep -Seconds $orbitMaxTimeToStartAndTeardown
          #Get-Service -Name $serviceName | %{ if ($_.Status -ne "Stopped") { throw "Fleet Service test #5 failed" } }

          # There is an sporadic issue with --insecure flag being used and osqueryd which causes long shutdown time, not testing this scenario until issue this scenario is sorted out

      - name: MSI Installer Tests
        shell: pwsh
        run: |
          # Tests setup
          $serviceName = "Fleet osquery"
          $registryPath = "HKLM:\SOFTWARE\FleetDM\"
          $installerExecTime = 15

          # Commenting test, being looked at as part of https://github.com/fleetdm/fleet/issues/8057

          # Test 1 - Check that there is not Orbit installation folder in programfiles and no registry entries after MSI uninstallation
          # msiexec /x ${{ steps.download.outputs.download-path }}\fleet-osquery.msi /quiet /passive /lv logtest1.txt
          # Start-Sleep -Seconds $installerExecTime
          # if (Test-Path -Path $Env:Programfiles\Orbit) { throw "MSI Installer test #1 failed" }
          # Get-Service -Name $serviceName -ErrorAction SilentlyContinue  | %{ if ($_.Name) { throw "MSI Installer test #1 failed" } }
          # if (((Get-ChildItem -Path $registryPath -ErrorAction SilentlyContinue | Measure-Object).Count) -gt 0) { throw "MSI Installer test #1 failed" }

          # Test 2 - Check that Orbit service, installation folder and registry entry are present after installing MSI again
          # msiexec /i ${{ steps.download.outputs.download-path }}\fleet-osquery.msi /quiet /passive /lv logtest2.txt
          # Start-Sleep -Seconds $installerExecTime
          # if (-not (Test-Path -Path $Env:Programfiles\Orbit)) { throw "MSI Installer test #2 failed" }
          # Get-Service -Name $serviceName -ErrorAction SilentlyContinue  | %{ if ($_.Status -ne "Running") { throw "MSI Installer test #2 failed" } }
          # if (((Get-ChildItem -Path $registryPath -ErrorAction SilentlyContinue | Measure-Object).Count) -eq 0) { throw "MSI Installer test #2 failed" }

          # Test 3 - Check that there is not Orbit folder in programfiles, no fleet service entry and no registry entries after uninstalling MSI again
          # msiexec /x ${{ steps.download.outputs.download-path }}\fleet-osquery.msi /quiet /passive /lv logtest3.txt
          # Start-Sleep -Seconds $installerExecTime
          # if (Test-Path -Path $Env:Programfiles\Orbit) { throw "MSI Installer test #3 failed" }
          # Get-Service -Name $serviceName -ErrorAction SilentlyContinue  | %{ if ($_.Name) { throw "MSI Installer test #3 failed" } }
          # if (((Get-ChildItem -Path $registryPath -ErrorAction SilentlyContinue | Measure-Object).Count) -gt 0) { throw "MSI Installer test #3 failed" }

          # Test 4 - Check that osquery manifest is present and that it points to the expected osqueryd.exe file
          # msiexec /i ${{ steps.download.outputs.download-path }}\fleet-osquery.msi /quiet /passive /lv logtest4.txt
          # Start-Sleep -Seconds $installerExecTime
          # Get-Content "$Env:Programfiles\Orbit\osquery.man" | % { if($_ -match 'resourceFileName=\"(.*?)\"') { if (-not (Test-Path -Path ([System.Environment]::ExpandEnvironmentVariables($Matches[1])))) { throw "MSI Installer test #4 failed" } } }

      - name: Upload Orbit logs
        if: always()
        uses: actions/upload-artifact@0b7f8abb1508181956e8e162db84b466c27e18ce # v2
        with:
          name: orbit-logs-windows
          path: C:\Windows\system32\config\systemprofile\AppData\Local\FleetDM\Orbit\Logs\orbit-osquery.log