Ephemeral JIT runner hangs after "Acknowledging runner request" and never exits; job fails with "lost communication" (2.338.0, linux-x64)
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 42/100
Research direction
Start with the runner listener logs and the request acknowledgement and credential token flow described in the report; compare the related issues #4309 and #2040 for clues. The report has no reproducer or named source files, so first trace where the listener waits after acknowledging a request and look for a way to reproduce the hang. Done means the runner starts the worker or exits cleanly when it cannot acquire the job.
Written by the indexing model from the issue text.
Description
Describe the bug
A runner registered with a JIT config (ephemeral, one job) accepts a job and then hangs. The listener's diag log ends at "Acknowledging runner request" and one token-request line. Nothing follows, and no Runner.Worker process is started. In the case we traced, GitHub failed the job about 10 minutes later with "The self-hosted runner lost communication with the server", and the registration was gone afterwards (GET /orgs/{org}/actions/runners/{id} returned 404).
The listener process never exits. It stays alive with 13 threads, its main thread in futex_wait and its TCP connections open, so the service manager never restarts it. Two runners on the same host hung this way about an hour apart. Each stayed hung for 6–7 hours, until we killed it.
To reproduce
We can't reproduce it on demand. Both cases had these conditions:
- Runner 2.338.0, linux-x64, Ubuntu 24.04, kernel 6.8.
- Registered via
generate-jitconfig..runnerhas"ephemeral": trueand"disableUpdate": true. Started throughrunsvc.shunder systemd, one job per start. - The job request arrived within about a second of "Listening for Jobs".
- The runner directory is read-only to the runner user, except
_workand_diag. In normal operation this causes only the expected "Read-only file system: .credentials" error after a completed job. Hundreds of jobs a day complete normally on this host.
Expected behavior
The job runs. Or, if the listener cannot acquire it, the listener gives up and exits so the supervisor can restart it.
Runner version / OS
2.338.0, Ubuntu 24.04.4 LTS x86_64
Logs
The listener diag log, last lines. Both instances end the same way:
[17:35:23Z INFO BrokerMessageListener] Session created.
[17:35:23Z INFO Terminal] WRITE LINE: Current runner version: '2.338.0'
[17:35:23Z INFO Terminal] WRITE LINE: 2026-11-11 17:35:23Z: Listening for Jobs
[17:35:23Z INFO JobDispatcher] Set runner/worker IPC timeout to 30 seconds.
[17:35:23Z INFO BrokerMessageListener] Acknowledging runner request '<id>'.
[17:35:23Z INFO CredentialManager] Creating credential type: OAuth
[17:35:23Z INFO RSAFileKeyManager] Loading RSA key parameters from file <runner>/.credentials_rsaparams
[17:35:23Z INFO GitHubActionsService] AAD Correlation ID for this token request: Unknown
Nothing follows these lines, and there is no Worker log.
Job annotation: "The self-hosted runner lost communication with the server. Verify the machine is running and has a healthy network connection. ..."
Possibly related: #4309, #2040.
- Dominant language
- C#
- Stars
- 6.4k
- Forks
- 1.5k
- Avg merge
- 1d 7h
- Merged PRs (30d)
- 36
Getting set up
Starts the project's dev container in your browser, under your own GitHub account.
- No Dockerfile or Docker Compose file
- No pull request template
- No contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from actions/runner
-
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
actions/runner#4820 · 3 comments ·
Maintainers usually reply within 1 day
-
Workflow schema omits documented code-quality and copilot-requests permissionsPossibly taken @localzet claimed this 12 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 1 day
-
Runner crashes at InitializeSecretMasker on machines with Thai (th-TH) culture — culture-sensitive LastIndexOf in PowerShellPreAmpersandEscapePossibly taken @CloudNinjaDev claimed this 47 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 88/100
Maintainers usually reply within 1 day
-
Broker retry logs an ERR stack trace and a phantom backoff on every job completion (listener's own cancellation treated as a request failure)Possibly taken @nitrobass24 claimed this 87 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 1 day
Similar issues
-
needs triage
Difficulty 1/5 Under an hour Newbie friendliness 71/100
Maintainers usually reply within 1 day
-
bug debugger flaky small
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
Maintainers usually reply within 1 day
-
[BUG] MyAnonaMouse search turns a curly apostrophe into a space, so titles with ’ are never foundOpen
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
Maintainers usually reply within 1 day
-
documentation
Difficulty 1/5 Under an hour Newbie friendliness 90/100
Maintainers usually reply within 1 day
-
About screenOpenenhancement
Difficulty 2/5 1-3 hours Newbie friendliness 65/100
GDMTT/JiraTaskBar#45 · 2 comments ·
Maintainers usually reply within 1 day