Runner listener exited and session duplicated

Describe the bug Runner listener exited with error code null and when it started, a duplicated session preventing the process to connect to the runner. Eventually our GHA job failed after 20 minutes plus

To Reproduce Steps to reproduce the behavior:

1. I’m not sure how to reproduce the issue. Probably we need to kill runner listener’s process
2. See error

Expected behavior Runner listener re-launch should resume the existing session or create a new session. If the runner is not functional it should not receive new job from GHA.

Workaround We delete the faulty runner pod, and has no issue with the new runner pod.

Runner Version and Platform

2.284.0

OS: Ubuntu 20.04/ Docker container in Kubernetes (AWS EKS 1.19)

What’s not working?

Runner pod’s logs

2021-11-23 17:57:50Z: Running job: terraform-cd
Runner listener exited with error code null
Runner listener exit with undefined return code, re-launch runner in 5 seconds.
Starting Runner listener with startup type: service
Started listener process, pid: 6021

√ Connected to GitHub

A session for this runner already exists.
2021-11-23 17:58:25Z: Runner connect error: The actions runner rr-devops-7c676ff546-dxzvx already has an active session for owner rr-devops-7c676ff546-dxzvx.. Retrying until reconnected.

[gitcos]

We’re having the same problem on a self-hosted runner for a GitHub Enterprise, and I found this issue with a Google search.

Here’s what we get in the runner’s logs, mildly edited to anonymize:

[2022-11-16 16:50:12Z ERR  GitHubActionsService] POST request to https://github.OURDOMAIN.com/_services/pipelines/[long hash deleted from public post]/_apis/distributedtask/pools/7/sessions failed. HTTP Status: Conflict
[2022-11-16 16:50:12Z ERR  MessageListener] Catch exception during create session.
[2022-11-16 16:50:12Z ERR  MessageListener] GitHub.DistributedTask.WebApi.TaskAgentSessionConflictException: The actions runner HOSTNAME-ARM64 already has an active session for owner HOSTNAME.

I can’t figure out which of the troubleshooting docs @nikola-jokic linked to above are relevant to this issue. None of them seem to be.

We’ve had this happen in the past, and the problem goes away after an hour - I assume this is because the existing session expires so GHE allows a new session to be created. But it seems like there should be a better process for this than leaving the runner offline for an hour.

P.S. I would reopen this issue, but I don’t seem to have the ability to (or I don’t know where to find that option in the UI).

[gitcos]

Just commenting to say we’ve run into this again a couple of times recently, and still haven’t found any better answer.