ramkrish86 commented on pull request #2501:
URL: https://github.com/apache/hbase/pull/2501#issuecomment-705490607


   `This is not batching to the NN? It is just doing archiving on new 
background thread.`
   Ya I did change the JIRA title. But not here. I will try to change the title 
if it allows. We are only doing it in separate thread now.
   
   `If we fail to complete an archiving because RS dies, will new RS pick up 
the archiving?`
   Actually speaking even if it fails it should be fine because if a WAL is not 
archived we can skip it while recovered edits are replayed assuming that the 
seqId has already been flushed to some hfile.
   BTW the entire 'abortable' case was brought in just to ensure the current 
behaviour. Since if the current Logroller thread trying to do the archive fails 
to do the archival we throw the Exception (from the AbstractFSWAL) and the log 
roller which has reference to the 'server' makes the 'abort' call. Since we are 
moving to a seperate thread and that AbstractFSWAL and its impl does not have 
access to 'server' we need to pass 'Abortable' every where. 


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


Reply via email to