sodonnel commented on PR #10440:
URL: https://github.com/apache/ozone/pull/10440#issuecomment-4741388441

   I scanned over this and my immediate thought it that there are a lot of 
places where allocation, deallocation and rollback have to happen - it would be 
easy for bugs to creep in.
   
   ReplicationManager tracks all its pending container creations and deletes in 
ContainerReplicaPendingOps - would it be cleaner and easier to hook into there 
to adjust the space allocated on a datanode? Once a container is added to the 
PendingOps tracking its definitely going to be sent - so we don't need to worry 
about the overloaded exceptions etc at that point or rollbacks.
   
   Pending ops is also cleared by the ICRs or a timeout if the new container 
does not appear in time. It feels like it would be a much more central place to 
add the accounting.


-- 
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]


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

Reply via email to