Logs can't be captured after some time for certain pods

[wenlinhe]

Describe the bug
Kubernetes env, using Promtail collecting pods logs and send to Loki and view the logs in Grafana.
Logs can't be captured after some time for certain pods. For these certain pods, there was logs shown in the beginning but then no new logs captured by Promtail, however there is new logs keep creating in the log file, the only workaround is the copy the log file to a new file then Promtail agent will read the log from new file, but this problem will repeat and repeat. There is no such problem for the other pods in the same Kubernetes(Openshift).

To Reproduce
Steps to reproduce the behavior:

1. Openshift v4.12+ env.
2. Promtail agent + kinds of applications
3. For a give application 'xxxx-jumpbox''s pods log, it can be shown in the beginning but then no logs collected
4. Try to copy the log to a new log file, Promtail can load the logs from new file

Screenshots, Promtail config, or terminal output