Anubhav-Roy commented on PR #20299: URL: https://github.com/apache/druid/pull/20299#issuecomment-6169386159
@FrankChen021 Thanks for taking the time to review again! As for the worker removal after snapshot is built, we still end up checking fresh at L:1158 before assigning it. The main purpose of this change is when we have a huge backlog of pending tasks, a sudden jump of tasks, into the queue. In the old implementation, it would end up locking for everyone and then keep computing which worker is free which ended up the cluster to stall. For the second comment, it's still less than the current implementation ( same comment as above). For a backlog of tasks, this is where the current implementation shows it perf improvement when we don't compute for each task. I'm sorry I didnt get the `WorkerHolder` approach. The problem we faced was this `statusLock` blocking for everyone. Could you please help clarify? Thanks! -- 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]
