mirror of
https://github.com/optim-enterprises-bv/kubernetes.git
synced 2025-12-04 15:15:36 +00:00
cefd95c51a
Because log capturing can end due to an error and because a pod has terminated, it is uncertain whether all log output has been captured. So far, the code leaned more towards restarting log capturing based on the rationale that duplicate logs are better than no logs. But this is confusing and potentially makes logs much larger, so now an additional heuristic is used to avoid log capturing when logging was started already and the pod itself is marked for deletion. That occurs before the individual containers shut down and get marked as terminated.