In some not well known scenarios, the docker execution fails and the runner does not update the job status correctly. This causes the job not to be marked as "finished" and it will be claimed again by the runner, resulting on a loop of executing a failing job. My suggestion is to add a claim_retries to jobs and a MAX_RETRIES so this failing jobs will not be claimed as they achieve this max.
I suggest a retries count of 2 or 3.
In some not well known scenarios, the docker execution fails and the runner does not update the job status correctly. This causes the job not to be marked as "finished" and it will be claimed again by the runner, resulting on a loop of executing a failing job. My suggestion is to add a
claim_retriesto jobs and aMAX_RETRIESso this failing jobs will not be claimed as they achieve this max.I suggest a retries count of 2 or 3.