Agent MSI support

.buildkite/pipeline.yml

@@ -2,22 +2,22 @@
 ### The following environment variables can be set for testing purposes via the buildkite UI or CLI
 ###  RUN_SNAPSHOT: "true" - will run the snapshot workflow
 ###  RUN_STAGING: "true" - will run the staging workflow
-###  DBRANCH: "main" - will override the dra branch param. Default is $BUILDKITE_BRANCH
+###  ONLY_AGENT: "true" - will build only the elastic-agent msi artifact
 ###
 
 steps:
-  - group: ":beats: Stack installers Snapshot"
+  - group: ":package: Stack installers Snapshot"
     key: "dra-snapshot"
     steps:
-      - label: ":hammer: Build stack installers Snapshot"
+      - label: ":construction_worker: Build stack installers / Snapshot"
         command: ".buildkite/scripts/build.ps1"
         key: "build-snapshot"
-        artifact_paths: "bin/out/**/*.msi"
+        artifact_paths: "c:/users/buildkite/esi/bin/out/**/*.msi"
         agents:
           provider: gcp
           image: family/ci-windows-2022
         env:
-          WORKFLOW: "snapshot"
+          DRA_WORKFLOW: "snapshot"
       - label: ":package: DRA Publish Snapshot"
         if: build.branch == 'main' || build.branch =~ /^[0-9]+\.[0-9]+\$/ || build.env("RUN_SNAPSHOT") == "true"
         command: ".buildkite/scripts/dra-publish.sh"
@@ -26,20 +26,20 @@ steps:
         agents:
           provider: gcp
         env:
-          WORKFLOW: "snapshot"
-  - group: ":beats: Stack installers Staging :beats:"
+          DRA_WORKFLOW: "snapshot"
+  - group: ":package: Stack installers Staging"
     key: "dra-staging"
     steps:
-      - label: ":hammer: Build stack installers staging"
+      - label: ":construction_worker: Build stack installers / Staging"
         if: build.branch =~ /^[0-9]+\.[0-9]+\$/ || build.env("RUN_STAGING") == "true"
         command: ".buildkite/scripts/build.ps1"
         key: "build-staging"
-        artifact_paths: "bin/out/**/*.msi"
+        artifact_paths: "c:/users/buildkite/esi/bin/out/**/*.msi"
         agents:
           provider: gcp
           image: family/ci-windows-2022
         env:
-          WORKFLOW: "staging"
+          DRA_WORKFLOW: "staging"
       - label: ":package: DRA Publish staging"
         if: build.branch =~ /^[0-9]+\.[0-9]+\$/ || build.env("RUN_STAGING") == "true"
         command: ".buildkite/scripts/dra-publish.sh"
@@ -48,7 +48,7 @@ steps:
         agents:
           provider: gcp
         env:
-          WORKFLOW: "staging"
+          DRA_WORKFLOW: "staging"
 
 notify:
   - slack: "#ingest-notifications"

.buildkite/scripts/build.ps1

@@ -1,72 +1,93 @@
+$ErrorActionPreference = "Stop"
+Set-Strictmode -version 3
+
+if (-not (Test-Path env:MANIFEST_URL)) {
+    $errorMessage = "Error: Required environment variable [MANIFEST_URL] is missing."
+    Write-Host $errorMessage
+    throw $errorMessage
+}
+
+# workaround path limitation for max 248 characters
+# example: https://buildkite.com/elastic/elastic-stack-installers/builds/3104#018c5e1b-23a7-4330-ad5d-4acc69157822/74-180
+cd ..
+# we can't use Rename-Item because this script runs from within the existing checkout resulting in
+# Rename-Item : The process cannot access the file because it is being used by another process.
+Copy-Item -Path .\elastic-stack-installers -Destination c:\users\buildkite\esi -Recurse
+cd c:\users\buildkite\esi
+
 # Read the stack version from build properties
 [xml]$xml = Get-Content -Path "Directory.Build.props"
 $ns = New-Object Xml.XmlNamespaceManager($xml.NameTable)
 $ns.AddNamespace("ns", "http://schemas.microsoft.com/developer/msbuild/2003")
 $stack_version = $xml.SelectSingleNode("//ns:PropertyGroup/ns:StackVersion", $ns).InnerText
-Write-Host "Building Stack version: $stack_version"
+$workflow = ${env:DRA_WORKFLOW}
+if ($workflow -eq "snapshot") {
+    $version = $stack_version + "-" + $workflow.ToUpper()
+} else {
+    $version = $stack_version
+}
 
-echo "~~~ Installing dotnet-sdk"
+Write-Host "~~~ Building Stack version: $stack_version"
+
+Write-Output "~~~ Installing dotnet-sdk"
 & "./tools/dotnet-install.ps1" -NoPath -JSonFile global.json -Architecture "x64" -InstallDir c:/dotnet-sdk
 ${env:PATH} = "c:\dotnet-sdk;" + ${env:PATH}
 Get-Command dotnet | Select-Object -ExpandProperty Definition
 
-echo "~~~ Reading msi certificate from vault"
-$MsiCertificate=& vault read -field=cert secret/ci/elastic-elastic-stack-installers/signing_cert
-$MsiPassword=& vault read -field=password secret/ci/elastic-elastic-stack-installers/signing_cert
-Remove-Item Env:VAULT_TOKEN
-
-$cert_home="C:/.cert"
-New-Item $cert_home -Type Directory -Force
-[IO.File]::WriteAllBytes("$cert_home/msi_certificate.p12", [Convert]::FromBase64String($MsiCertificate))
-[IO.File]::WriteAllText("$cert_home/msi_password.txt", $MsiPassword)
-echo "Certificate successfully written to $cert_home"
-
 $client = new-object System.Net.WebClient
 $currentDir = $(Get-Location).Path
 $beats = @('auditbeat', 'filebeat', 'heartbeat', 'metricbeat', 'packetbeat', 'winlogbeat')
 $ossBeats = $beats | ForEach-Object { $_ + "-oss" }
-$workflow = ${env:WORKFLOW}
 
-echo "~~~ downloading beat $workflow dependencies"
+# TODO remove (and all references/conditionals below) after testing; this just helps to speed up elastic-agent specific build tests
+$onlyAgent = $env:ONLY_AGENT -as [string]
+
 Remove-Item bin/in -Recurse -Force -ErrorAction Ignore
 New-Item bin/in -Type Directory -Force
-if ($workflow -eq "snapshot") {
-    $version = $stack_version + "-" + $workflow.ToUpper()
-    $hostname = "artifacts-snapshot.elastic.co"
-    $response = Invoke-WebRequest -UseBasicParsing -Uri "https://$hostname/beats/latest/$version.json"
-    $json = $response.Content | ConvertFrom-Json
-    $buildId = $json.build_id
-    $prefix = "$hostname/beats/$buildId"
-} else {
-    $version = $stack_version
-    $hostname = "artifacts-staging.elastic.co"
-    $response = Invoke-WebRequest -UseBasicParsing -Uri "https://$hostname/beats/latest/$version.json"
-    $json = $response.Content | ConvertFrom-Json
-    $buildId = $json.build_id
-    $prefix = "$hostname/beats/$buildId"
-}
-foreach ($beat in ($beats + $ossBeats)) {
-    try {
-        $beatName = $beat.Replace("-oss", "")
-        $url = "https://$prefix/downloads/beats/$beatName/$beat-$version-windows-x86_64.zip"
-        echo "Downloading from $url"
+
+$manifestUrl = ${env:MANIFEST_URL}
+$response = Invoke-WebRequest -UseBasicParsing -Uri $manifestUrl
+$json = $response.Content | ConvertFrom-Json
+
+Write-Output "~~~ Downloading $workflow dependencies"
+
+$urls = @()
+
+try {
+    $packageName = "elastic-agent"
+    $projectName = "$packageName-package"
+    $urls += $json.projects.$projectName.packages."$packageName-$version-windows-x86_64.zip".url
+
+    if ($onlyAgent -eq "true") {
+        Write-Output "Skipping beats because env var ONLY_AGENT is set to [$env:ONLY_AGENT]"
+    }
+    else {
+        $projectName = "beats"
+        foreach ($packageName in ($beats + $ossBeats)) {
+            $urls += $json.projects.$projectName.packages."$packageName-$version-windows-x86_64.zip".url
+        }
+    }
+    foreach ($url in $urls) {
+        $destFile = [System.IO.Path]::GetFileName($url)
+        Write-Output "Downloading from $url to $currentDir/bin/in/$destFile"
         $client.DownloadFile(
             $url,
-            "$currentDir/bin/in/$beat-$version-windows-x86_64.zip"
+            "$currentDir\bin\in\$destFile"
         )
     }
-    catch [System.Net.WebException] {
-        if ($_.Exception.InnerException) {
-            Write-Error $_.Exception.InnerException.Message
-        } else {
-            Write-Error $_.Exception.Message
-        }
-        throw "An error was encountered while downloading dependencies, aborting."
+} catch [System.Net.WebException] {
+    if ($_.Exception.InnerException) {
+        Write-Error $_.Exception.InnerException.Message
+    } else {
+        Write-Error $_.Exception.Message
     }
+    throw "An error was encountered while downloading dependencies, aborting."
 }
 
-echo "--- Building $workflow msi"
-$args = @(
+Remove-Item bin/out -Recurse -Force -ErrorAction Ignore
+
+Write-Output "--- Building $workflow msi"
+$cliArgs = @(
     "run",
     "--project",
     "src\build\ElastiBuild.csproj",
@@ -75,25 +96,41 @@ $args = @(
     "--",
     "build",
     "--cid",
-    $version,
-    "--cert-file",
-    "$cert_home/msi_certificate.p12",
-    "--cert-pass",
-    "$cert_home/msi_password.txt"
+    $version
 )
-$args += ($beats + $ossBeats)
-&dotnet $args
+$cliArgs += "elastic-agent"
+if ($onlyAgent -ne "true") {
+    $cliArgs += ($beats + $ossBeats)
+}
+
+&dotnet $cliArgs
 if ($LastExitcode -ne 0) {
-    Write-Error "Build$workflow failed with exit code $LastExitcode"
+    Write-Error "Build $workflow failed with exit code $LastExitcode"
     exit $LastExitcode
 } else {
-    echo "Build$workflow completed with exit code $LastExitcode"
+    Write-Output "Build $workflow completed with exit code $LastExitcode"
 }
 
-echo "--- Checking that all artefacts are there"
+Write-Output "--- Checking that all artifacts are there"
 $msiCount = Get-ChildItem bin/out -Include "*.msi" -Recurse | Measure-Object | Select-Object -ExpandProperty Count
-$expected = 2 * $beats.Length
+
+$expected = 1
+if ($onlyAgent -ne "true") {
+    $expected += (2 * $beats.Length)
+}
+
 if ($msiCount -ne $expected) {
-    Write-Error "Expected $expected msi executable to be produced, but $msiCount were"
+    Write-Error "Failed: expected $expected msi executables to be produced, but $msiCount were found."
     exit 1
+} else {
+    Write-Output "Success, found $msiCount artifacts in bin/out."
 }
+
+try {
+    & (Join-Path $PSScriptRoot "test.ps1")
+    write-host "Testing Completed"
+} catch {
+    write-host "Testing Failed"
+    write-error $_
+    exit 1
+}

.buildkite/scripts/test.ps1

@@ -0,0 +1,45 @@
+$ErrorActionPreference = "Stop"
+Set-Strictmode -version 3
+
+write-host (ConvertTo-Json $PSVersiontable -Compress)
+write-host "Running as: $([Environment]::UserName)"
+
+write-host "`$env:AGENT = $($Env:AGENT)"
+
+if ($psversiontable.psversion -lt "7.4.0") {
+    # Download Powershell Core, and rerun this script using Powershell Core
+
+    write-host "Downloading Powershell Core"
+    invoke-webrequest -uri https://github.com/PowerShell/PowerShell/releases/download/v7.4.0/PowerShell-7.4.0-win-x64.zip -outfile pwsh.zip
+
+    write-host "Expanding Powershell Core"
+    Expand-Archive pwsh.zip -destinationpath (Join-Path $PSScriptRoot "pwsh")
+
+    Write-host "Invoking from Powershell Core"
+    & (Join-Path $PSScriptRoot "pwsh/pwsh.exe") -file (Join-Path $PSScriptRoot "test.ps1")
+
+    if ($LASTEXITCODE -eq 0) {
+        write-host "Child pwsh process exited successfully"
+        exit 0
+    } else {
+        write-host "Child pwsh process returned $LASTEXITCODE, a non zero exit code"
+        throw "Tests failed."
+    }
+}
+
+$AgentMSI = Get-ChildItem bin/out -Include "elastic-agent*.msi" -Recurse
+
+if ($AgentMSI -eq $null) {
+    write-error "No agent MSI found to test"
+}
+
+
+$OldAgentMSI = (Join-Path $PSScriptRoot "elastic-agent-8.11.4-windows-x86_64.msi")
+if (-not (test-path $OldAgentMSI)) {
+    Write-Host "Downloading older MSI for upgrade tests"
+    invoke-webrequest -uri https://storage.googleapis.com/agent-msi-testing/elastic-agent-8.11.4-windows-x86_64.msi -outfile $OldAgentMSI
+}
+
+& (Join-Path $PSScriptRoot "../../src/agent-qa/Invoke-Pester.ps1") -PathToLatestMSI $AgentMSI.Fullname -PathToEarlyMSI $OldAgentMSI
+
+write-host "Returned from Pester Test"

.gitignore

@@ -63,4 +63,6 @@ src/*.cmd
 # Executable Artifacts
 src/*.exe
 src/*.pdb
-src/*.dll
+src/*.dll
+pwsh
+pwsh.zip

README.md

@@ -43,3 +43,28 @@ Update version in `Directory.Build.props` in the branch for the related minor ve
       - Add a new daily schedule for the new minor branch
       - Remove the daily schedule for the previous minor branch
     ex: https://github.com/elastic/elastic-stack-installers/pull/156 and https://github.com/elastic/elastic-stack-installers/pull/172
+
+---
+## Agent
+
+In case of problems during install / uninstall of agent, please refer to the [Capturing Logs](https://github.com/elastic/elastic-stack-installers/blob/agent_support/README.md#capturing-logs) section which will enable troubleshooting.
+
+### Install
+During the install flow, The MSI installer will unpack the contents of the MSI to a temp folder and then will call the `elastic-agent install` in order to:
+1. copy the files to the final destination at `c:\Program Files\Elastic\Agent`
+2. register the agent as a windows service
+3. enroll the agent into fleet
+
+In order to complete step 3 above, the MSI installer shall receive command line arguments, passed with INSTALLARGS command line switch followed by `"`, for example:
+```
+elastic-agent.msi INSTALLARGS="--url=<fleet_url_with_port> --enrollment-token=<token>"
+```
+
+Note that the MSI will call the `elastic-agent install` command with `-f` (force) to avoid user interaction.
+
+### Uninstall
+Similarly to the install flow (described above), the MSI will call the `elastic-agent uninstall` command, and it's possible to pass arguments using `INSTALLARGS`. One common use case is uninstalling an agent which has tamper protection enabled.
+
+### Upgrade
+The Agent MSI doesn't support upgrade. Since the agents are fleet managed, upgrades shall be done using fleet (UI / API).
+

catalog-info.yaml

@@ -53,11 +53,6 @@ spec:
         publish_commit_status: true
         publish_commit_status_per_step: false
       repository: elastic/elastic-stack-installers
-      schedules:
-        Daily main:
-          branch: main
-          cronline: "@daily"
-          message: Builds daily `main` stack-installers dra
       teams:
         everyone:
           access_level: BUILD_AND_READ
@@ -93,15 +88,8 @@ spec:
         publish_commit_status_per_step: false
       repository: elastic/elastic-stack-installers
       schedules:
-        Daily 8_12:
-          branch: "8.12"
-          cronline: "*/10 * * * *"
-          message: Checking for new beats artifacts for `8.12`
-        Daily 8_11:
-          branch: "8.11"
-          cronline: "*/10 * * * *"
-          message: Checking for new beats artifacts for `8.11`
-        Weekly main:
+        # trigger for 7_17 still needed for now, details in https://elasticco.atlassian.net/browse/REL-1004?focusedCommentId=107973
+        Daily 7_17:
           branch: "7.17"
           cronline: "*/10 * * * *"
           message: Checking for new beats artifacts for `7.17`

src/CustomAction.config

@@ -0,0 +1,11 @@
+<?xml version="1.0" encoding="utf-8" ?>
+                                                <configuration>
+                                                    <startup useLegacyV2RuntimeActivationPolicy="true">
+                                                        <supportedRuntime version="v4.0.30319"/>
+                                                        <supportedRuntime version="v4.0" sku=".NETFramework,Version=v4.5"/>
+                                                        <supportedRuntime version="v4.0" sku=".NETFramework,Version=v4.0"/>
+                                                        <supportedRuntime version="v2.0.50727"/>
+                                                        <supportedRuntime version="v2.0.50215"/>
+                                                        <supportedRuntime version="v1.1.4322"/>
+                                                    </startup>
+                                                </configuration>