huyuanfeng2018 opened a new issue, #4998:
URL: https://github.com/apache/paimon/issues/4998

   ### Search before asking
   
   - [x] I searched in the [issues](https://github.com/apache/paimon/issues) 
and found nothing similar.
   
   
   ### Paimon version
   
   0.8+
   
   ### Compute Engine
   
   Flink
   
   ### Minimal reproduce step
   
   1. Use cdc action and do not configure checkpoint intervals (Paimon will 
actively configure checkpoint via: #2461)
   2. open  `commit.force-create-snapshot` = `true`
   3. Actively kill a taskmanager during a task run, triggering a flink 
failover.
   
   
   
   
   
   ### What doesn't meet your expectations?
   
   ## Exception
   Chances are, we'll find that the task keeps restarting.
   
   
   ## How this led to this Exception
   
   - With `commit.force-create-snapshot` turned on, a commit will be triggered 
to generate a snapshot of paimon even if no data is written.
   
   - flink's failover resumes from checkpoint, based on paimon's two-phase 
commit, and resuming from checkpoint may trigger a commit of the manifest file
   
   - If there are no new data files in the commit (via the 
streamingCheckpointEnabled variable in the `CommitterOperator`), the manifest 
file commit may be filtered out because we didn't actively set the checkpoint. 
`streamingCheckpointEnabled= false`, then the commit will be ignored (because 
no new data file has been generated). Then, after the second job restart, the 
process will be repeated again, and we will be stuck in a dead loop.
   
   
   
   
   ### Anything else?
   
   _No response_
   
   ### Are you willing to submit a PR?
   
   - [x] I'm willing to submit a PR!


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to