HeartSaVioR commented on pull request #28412:
URL: https://github.com/apache/spark/pull/28412#issuecomment-655125113


   @baohe-zhang 
   Thanks for the update. This is really helpful. So the small event log file 
shows there's a chance the ratio can be beyond 1/2.
   
   Personally it seems OK to simply allow such case, as these small event logs 
can be migrated to the LevelDB KV store fast enough, and the memory usage would 
hold shortly. That said, it'd be simpler if we can just go through with LevelDB 
KV store for small event logs, no need to run background thread per app log.
   
   Either is fine for me. For latter the lower bound should be configured 
against uncompressed log, and we should do the similar estimation on compressed 
log before applying. That's another tricky one, but probably good to have. (I 
wouldn't run background thread for event log which can be loaded in a couple of 
seconds)
   
   @tgravescs WDYT?


----------------------------------------------------------------
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]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to