StephanEwen commented on issue #10768: [FLINK-15012][checkpoint] Introduce 
shutdown to CheckpointStorageCoordinatorView to clean up checkpoint directory.
URL: https://github.com/apache/flink/pull/10768#issuecomment-594422896
 
 
   Should we make the directory cleanup **NON recursive**? That might be the 
safer option.
   
   Given that we only want the parent/container directories removed, 
non-recursive delete should be fine.
   
   If something is still in those directories, it may be unsafe to delete them. 
Maybe a retained checkpoint from a previous job? Maybe an incremental safepoint 
in the future? It may be dangerous to assume that at shutdown time, all data in 
these directories can be deleted.
   If we see left-over data because checkpoints failed to clean up properly, 
then the shutdown part may not be the right place to clean this up, because it 
has too little information to know if this truly safe to be deleted. For that 
we need some "cleanup safety nets" that understand which shared files are still 
referenced, which checkpoints have been discarded (up to which point).

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to