UI Jobs streams are not cleaned up after PR is closed

[quynhnhatnguyen]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request. Searching for pre-existing feature requests helps us consolidate datapoints for identical requirements into a single place, thank you!
• Please do not leave "+1" or other comments that do not add relevant new information or questions, they generate extra noise for issue followers and do not help prioritize the request.
• If you are interested in working on this issue or have submitted a pull request, please leave a comment.

---

Overview of the Issue

Atlantis UI Jobs Stream are not cleaned up after a PR is closed, resulted in system resource are not freed up.

Reproduction Steps

1. Open a PR in Github, plan or apply the PR to generate jobs stream. Visit the job's stream to confirm seeing the stream.
2. Close the PR in Github. Visit the UI to see that jobs are still listed whereas locks are cleaned. Confirm that data in repos dir is also cleaned up fine. Visit the previous job stream URL to confirm that you still can see the streams.

Environment details

• Atlantis version: 0.33.0
• Deployment method: GKE

Additional Context

Screenshots of before closing the PR:

Screenshots of after closing the PR:

• Locks are gone but streams remains in the table and you can also click to view the stream content)

• System resource is not freed up after closing a PR due to streams are left behind

[dosubot]

Hey @quynhnhatnguyen 👋, I'm here to help you with any bugs, questions, or contributions. Let's tackle this issue together!

The issue you're experiencing, where UI job streams are not cleaned up after a PR is closed, seems to be related to the cleanup process managed by the PullClosedExecutor in Atlantis. This component is responsible for cleaning up resources, including log streaming resources, when a PR is closed ^[1].

The expected behavior is that log streaming resources should be cleaned up, workspaces deleted, and locks removed when a PR is closed ^[2]. If this is not happening, it might indicate a bug or misconfiguration in the cleanup process.

To address this issue, you might want to:

1. Verify that the PullClosedExecutor is correctly configured and that all its dependencies (like LogStreamResourceCleaner, Locker, and WorkingDir) are functioning as expected.
2. Check for any recent changes or updates in the codebase that might have affected the cleanup process.
3. Ensure that your Atlantis version (0.33.0) is up-to-date with any patches or fixes related to this issue.

If the problem persists, it might be helpful to look into the specific implementation details or logs to identify any errors during the cleanup process.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[joe1981al]

@quynhnhatnguyen, can you check if those logs are complete? We are seeing logs being persistent as well but when clicked the displayed log is truncated. If the log page is refreshed it is truncated at a different spot during each refresh. The full log was posted in the comment correctly.

Running atlantis:v0.33.0 container deployed via terraform-gce-atlantis?ref=v4.2.0

[quynhnhatnguyen]

@joe1981al ours is not truncated. The full log is persistent in the streams of closed PRs in my test.

[joe1981al]

@quynhnhatnguyen my live log issue is unrelated, submitted issue to gce-terraform-atlantis for that.

Which git repo are you using? We are using Azure DevOps and seeing similar behavior. I'm not familiar with golang but took a stab at debugging.

Seeing a behavior here that seems unexpected. It's not matching in the if statement and skips the cleanup steps.

Added some logging and p.pullToJobMapping contains content like this:

{
 "{54106 repo azureorg/project/repo  terraform/project/path default}": {
  "0cea6de8-5618-405f-b819-42d377b6f5ae": {
   "JobID": "0cea6de8-5618-405f-b819-42d377b6f5ae",
   "JobIDUrl": "",
   "JobDescription": "",
   "Time": "2025-02-21T12:44:55.947183432-06:00",
   "TimeFormatted": "",
   "JobStep": "plan"
  }
 }
}

but pullInfo contains

jobs.PullInfo {PullNum: 54106, Repo: "repo", RepoFullName: "", ProjectName: "", Path: "", Workspace: "default"}

With this, I assume the load fails to find the job id which causes the failure. Going to see if I can find where pullinfo is populated and see how hard it is to update.

[joe1981al]

Believe I have it figured out, have it working locally.

This area, I believe, adds entries to the plan log and has several inputs for pullinfo that are not populated in this section when a cleanup is called. If updated as below, it succeeds in deleting the plan log and entries disappear from UI.

	if pullStatus != nil {
		for _, project := range pullStatus.Projects {
			jobContext := jobs.PullInfo{
				PullNum:      pull.Num,
				Repo:         pull.BaseRepo.Name,
				RepoFullName: pull.BaseRepo.FullName,
				ProjectName:  project.ProjectName,
				Path:         project.RepoRelDir,
				Workspace:    project.Workspace,
			}
			p.LogStreamResourceCleaner.CleanUp(jobContext)
		}
	}

[joe1981al]

Issue has likely existed for over a year since this commit added more fields to PullInfo but did not update all things that use it to include those new fields.

[quynhnhatnguyen]

@joe1981al Thanks for the quick confirmation and patch.

Do you want to make a PR for this fix?