Today at 9 AM (UTC) a job has finished on one of my own runners. The job coordinator started responding with 503 codes and caused my runner to hang since then.
Here’s one of the log lines from journalctl:
Nov 05 09:21:53 rendernode-0 gitlab-runner[15102]: WARNING: Submitting job to coordinator... failed code=503 job=341715017 job-status= runner=B9fZJehc status=503 Service Unavailable
This line is repeated ten times, twice for each timestamp, with 3-4 seconds between timestamps.