One of our Broker disks ran out of space and went offline. The broker is back online but it appears 2 of the checkpoint files got stuck while out of space, and could not rollover the tmp files:
Should I manually rename the tmp files or is there another recommended action?
did you get any errors during startup regarding the tmp files?
did you check the file size?
some insights regarding the files
Thanks for the info.
I did miss a startup error for this disk:
Malformed line in checkpoint file (/hadoop/kafka/kafka-logs/log-start-offset-checkpoint)
The log-start-offset-checkpoint does look malformed with many “@@@@” and no text.
log-start-offset-checkpoint.tmp looks valid.
hmm ok I see
without seeing everything I would try to rename the problematic file
and start over again.