You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by GitBox <gi...@apache.org> on 2021/05/24 10:03:30 UTC

[GitHub] [ozone] sodonnel commented on pull request #2268: HDDS-5249. Race Condition between Full and Incremental Container Reports

sodonnel commented on pull request #2268:
URL: https://github.com/apache/ozone/pull/2268#issuecomment-846933140


   @GlenGeng Thanks for the review. I will go ahead and commit this one, and raise a new Jira for issue 2.
   
   I think you are correct, in that we may need a monotonically increasing sequence number in the datanode, plus maybe some locking in the DN to guarantee the contents of the FCR.
   
   The hard part, is figuring out what to do with that information on SCM. I wonder if we need to attach the "report sequence number" to every replica in SCM. Then we can say:
   
   * The FCR is has sequence 100, and it does not have container 1001.
   * 1001 is in SCM, but added by ICR sequence 101.
   * Right now, the FCR processing would remove container 1001, but with the sequence we could ensure it does not remove anything created by a newer sequence number.
   
   However this approach adds a new field to every replica which would have a memory overhead, so we need to try to think of a better way.


-- 
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:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org