The successful and failed job history limits work to retain jobs.
However on a large busy cluster, while the jobs are retained, the pods for those jobs are deleted, losing the logs.
OTOH on a small dev cluster, both the jobs and their pods are retained. Not sure why but setting the ttlSecondsAfterFinished to a large value (90 days by default) could help.
The successful and failed job history limits work to retain jobs. However on a large busy cluster, while the jobs are retained, the pods for those jobs are deleted, losing the logs. OTOH on a small dev cluster, both the jobs and their pods are retained. Not sure why but setting the ttlSecondsAfterFinished to a large value (90 days by default) could help.