Refactor to optimise for Windows service manager checkin #9672

michalpristas · 2025-09-01T12:17:41Z

This PR is huge but it does just few things.

It does not use elastic-agent-libs for calling ProcessWindowsControlEvents. ProcessWindowsControlEvents creates an objects that serves as an service instance and communicates with service manager. But having this in libs has an effect of running init section of whole elastic-agent-libs dependency tree before registering with a service.

It splits agent/cmd package into sub packages isolating each command. Service Manager communication is then moved outside to internal/pkg/agent/agentservice as described in first step. Having this split and service in a package deliberately named makes init section of agentservice being called sooner during initialization.

In agentservice init service we add WaitGroup when spinning up ProcessWindowsControlEvents goroutine. This blocks loading of subsequent packages and avoids possibility of starving ourselves of resources (when new goroutines are started later, subprocesses...). We cannot guarantee ordering of goroutines and that this one will be up in time. This is best effort of achieving that.
It's essential to have a package named like this because of the way how init section loading is implemented (alphabetical sorting plays a significant role)

Result of this is that we not just moved communication a way sooner to init section (normally end of init due to big dependency tree)
But also moved it before proceeding with initialization of

composables
azure, aws sdks,
k8s,
prometheus. and cockroachdb,
>95% of otel dependencies,
beats

This makes separation of otel into custom binary not critical for issues related to windows service manager communication timeouts.

To make it easier I added comment with a code change as these are not that frequent

Related #4971

elasticmachine · 2025-09-02T10:25:51Z

Pinging @elastic/elastic-agent-control-plane (Team:Elastic-Agent-Control-Plane)

…t-svc-start

mergify · 2025-09-05T12:09:00Z

This pull request is now in conflicts. Could you fix it? 🙏
To fixup this pull request, you can check out it locally. See documentation: https://help.github.com/articles/checking-out-pull-requests-locally/

git fetch upstream
git checkout -b play/init-svc-start upstream/play/init-svc-start
git merge upstream/main
git push upstream play/init-svc-start

internal/pkg/agent/agentservice/agentservice_windows.go

Co-authored-by: Ben McNichols <mcnichols@gmail.com>

…c-agent into play/init-svc-start

mergify · 2025-09-10T06:56:37Z

This pull request is now in conflicts. Could you fix it? 🙏
To fixup this pull request, you can check out it locally. See documentation: https://help.github.com/articles/checking-out-pull-requests-locally/

git fetch upstream
git checkout -b play/init-svc-start upstream/play/init-svc-start
git merge upstream/main
git push upstream play/init-svc-start

michalpristas · 2025-09-05T07:20:04Z

internal/pkg/agent/agentservice/agentservice_windows.go

+// couldNotConnect is the errno for ERROR_FAILED_SERVICE_CONTROLLER_CONNECT.
+const couldNotConnect syscall.Errno = 1063
+
+type beatService struct {


code change: bringin in dependency

michalpristas · 2025-09-05T07:20:20Z

internal/pkg/agent/cmd/agentrun/run.go

+	// After this is run, the service is considered by the OS to be stopped.
+	// This must be the first deferred cleanup task (last to execute).
+	defer func() {
+		agentservice.NotifyTermination()


code change: calling internal implementation

michalpristas · 2025-09-05T07:20:38Z

internal/pkg/agent/cmd/agentrun/run.go

+		agentservice.WaitExecutionDone()
+	}()
+
+	service.BeforeRun()


these are original

michalpristas · 2025-09-05T07:21:03Z

internal/pkg/agent/cmd/agentrun/run.go

+// or more contributor license agreements. Licensed under the Elastic License 2.0;
+// you may not use this file except in compliance with the Elastic License 2.0.
+
+package agentrun


code change: run -> agentrun to help with alphabetical ordering during init

michalpristas · 2025-09-05T07:21:47Z

testing/integration/ess/init_test.go

+	"github.com/elastic/elastic-agent/testing/integration"
+)
+
+func TestInitOrderNotDegraded(t *testing.T) {


code change: added test
not entirely happy with testing criteriabut at least it checks it is called during init and not at the end of it

internal/pkg/agent/agentservice/agentservice_windows.go

…t-svc-start

mergify · 2025-09-26T09:34:49Z

This pull request is now in conflicts. Could you fix it? 🙏
To fixup this pull request, you can check out it locally. See documentation: https://help.github.com/articles/checking-out-pull-requests-locally/

git fetch upstream
git checkout -b play/init-svc-start upstream/play/init-svc-start
git merge upstream/main
git push upstream play/init-svc-start

…t-svc-start

elastic-sonarqube · 2025-09-26T12:50:40Z

Quality Gate failed

Failed conditions
21.1% Coverage on New Code (required ≥ 40%)

See analysis details on SonarQube

elasticmachine · 2025-09-26T13:14:14Z

💛 Build succeeded, but was flaky

Buildkite Build
Commit: 29fb35c

Failed CI Steps

History

💔 Build #27584 failed 15310b3
💚 Build #26452 succeeded 78b296f
💔 Build #26451 failed e6c2b66
💔 Build #26371 failed df768a4
💚 Build #26139 succeeded af57714

cc @michalpristas

mergify · 2025-10-01T07:00:34Z

This pull request is now in conflicts. Could you fix it? 🙏
To fixup this pull request, you can check out it locally. See documentation: https://help.github.com/articles/checking-out-pull-requests-locally/

git fetch upstream
git checkout -b play/init-svc-start upstream/play/init-svc-start
git merge upstream/main
git push upstream play/init-svc-start

ebeahan · 2025-11-05T19:25:59Z

@michalpristas do we still need to pursue the changes here? Or can we close?

michalpristas · 2025-11-05T20:02:07Z

My preference would be to close and go with Blakes proposal as it is less fragile, more deterministic

michalpristas added 3 commits August 29, 2025 11:54

Moved processing of windows events to init

df03819

cmd refactor

9878a56

conflicts

736f968

michalpristas self-assigned this Sep 1, 2025

michalpristas added bug Something isn't working Team:Elastic-Agent-Control-Plane Label for the Agent Control Plane team backport-skip skip-changelog labels Sep 1, 2025

michalpristas added 8 commits September 1, 2025 14:51

linux files

ef25c7a

uid,gid related linter

93b7e61

interactive

ccc6a66

headers

1ef74cd

notice

0cc8cb3

move testdata

e9e04ac

ess tests compilation errors

e1fc526

Merge branch 'main' into play/init-svc-start

af2ab1a

michalpristas marked this pull request as ready for review September 2, 2025 10:25

michalpristas requested a review from a team as a code owner September 2, 2025 10:25

michalpristas requested review from blakerouse and nkvoll September 2, 2025 10:25

michalpristas requested a review from cmacknz September 2, 2025 10:26

ebeahan removed the request for review from nkvoll September 2, 2025 15:42

michalpristas added backport-active-all Automated backport with mergify to all the active branches and removed backport-skip labels Sep 3, 2025

michalpristas added 2 commits September 3, 2025 12:10

fixed termination and logging

ec8f508

Merge branch 'main' of github.com:elastic/elastic-agent into play/ini…

85806db

…t-svc-start

michalpristas marked this pull request as draft September 3, 2025 13:46

try init test

a93cab0

michalpristas force-pushed the play/init-svc-start branch from 05457bc to a93cab0 Compare September 3, 2025 14:16

michalpristas added 2 commits September 4, 2025 11:29

output already captured

b858dd3

output already captured

af57714

michalpristas marked this pull request as ready for review September 4, 2025 11:43

michalpristas added 6 commits September 8, 2025 16:32

conflicts with disk space errors refactor

6614a9e

conflicts with disk space errors refactor

4f30d0c

conflicts with disk space errors refactor

4dcabe4

conflicts with disk space errors refactor

0cec2b0

use SetupMetricsOptions

fc73811

remove unused param

df768a4

bjmcnic reviewed Sep 8, 2025

View reviewed changes

michalpristas and others added 4 commits September 9, 2025 09:59

Update internal/pkg/agent/agentservice/agentservice_windows.go

e6c2b66

Co-authored-by: Ben McNichols <mcnichols@gmail.com>

updated comment

2232721

Merge branch 'play/init-svc-start' of github.com:michalpristas/elasti…

6fb5afc

…c-agent into play/init-svc-start

goimports

78b296f

michalpristas commented Sep 25, 2025

View reviewed changes

michalpristas added 3 commits September 26, 2025 09:42

resolved conflicts

f6ce8ee

increased timout to prevent premature cancellation

f161b82

Merge branch 'main' of github.com:elastic/elastic-agent into play/ini…

15310b3

…t-svc-start

michalpristas added 2 commits September 26, 2025 12:59

fixed path

0979606

Merge branch 'main' of github.com:elastic/elastic-agent into play/ini…

29fb35c

…t-svc-start

cmacknz mentioned this pull request Sep 26, 2025

Extend use of the Windows archive proxy to elastic-agent.exe to runtime to register with the Windows service manager as fast as possible #10183

Open

ebeahan closed this Nov 6, 2025

Refactor to optimise for Windows service manager checkin #9672

Refactor to optimise for Windows service manager checkin #9672

Uh oh!

Conversation

michalpristas commented Sep 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

elasticmachine commented Sep 2, 2025

Uh oh!

mergify bot commented Sep 5, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mergify bot commented Sep 10, 2025

Uh oh!

michalpristas Sep 5, 2025

Choose a reason for hiding this comment

Uh oh!

michalpristas Sep 5, 2025

Choose a reason for hiding this comment

Uh oh!

michalpristas Sep 5, 2025

Choose a reason for hiding this comment

Uh oh!

michalpristas Sep 5, 2025

Choose a reason for hiding this comment

Uh oh!

michalpristas Sep 5, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mergify bot commented Sep 26, 2025

Uh oh!

elastic-sonarqube bot commented Sep 26, 2025

Quality Gate failed

Uh oh!

elasticmachine commented Sep 26, 2025

💛 Build succeeded, but was flaky

Failed CI Steps

History

Uh oh!

mergify bot commented Oct 1, 2025

Uh oh!

ebeahan commented Nov 5, 2025

Uh oh!

michalpristas commented Nov 5, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

michalpristas commented Sep 1, 2025 •

edited

Loading