jihoonson commented on pull request #11243: URL: https://github.com/apache/druid/pull/11243#issuecomment-840128738
@capistrant yes, because it can be useful for the replacing rolling downgrade. In the replacing rolling downgrade, you can terminate nodes one at a time and add a new node that replaces with the terminated one. Suppose you have your cluster upgraded and a task running that uses the new segment allocation protocol. But a new critical bug is found in the new version and now you want to downgrade your cluster. In this case, you first need to wait until there is no running task that uses the new protocol before you start the downgrade process because every task must use the legacy protocol during the downgrade. To do this, you will need to either downgrade the overlord first or set `useLineageBasedSegmentAllocation` to false in the default taskContext. If you cannot downgrade the overlord first for some reason, you should be able to set it to false. -- 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: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@druid.apache.org For additional commands, e-mail: commits-h...@druid.apache.org
