"Runner listener exit with retryable error" breaks ephemeral behaviour

[fkorotkov]

Describe the bug

We are running ephemeral runners and sometimes experiencing the following behaviour:

• Runner successfully connects and start listening for jobs indicated by Listening for Jobs log message.
• After around two minutes of listening Runner listener exit with retryable error, re-launch runner in 5 seconds. printed out
• Worker restarts but is getting Failed to create a session. The runner registration has been deleted from the server, please re-configure.

It seems runner re-register itself before restarting but is not renewing registration.

To Reproduce

I guess you need constantly run it.

Expected behavior

Runners reconnects without unregistering.

Runner Version and Platform

Arm64 on Apple Silicon

What's not working?

see above

Job Log Output

not relevant

Runner and Worker's Diagnostic Logs

We are not collecting these logs from the ephemeral VMs.