jiajunwang commented on issue #680: Question on CrushEd rebalance algorithm and MasterSalve state model for Pinot URL: https://github.com/apache/helix/issues/680#issuecomment-573936705 Hi @kishoreg, I agree with the first point, that's why I think there is no problem to switch for now. Regarding the OnlineOffline state model, Helix will try to bring all the replicas to the topstate "online". But what Pinot needs is only one topstate replica, right? That is the thing in my mind. Please correct me if I misunderstand the usage. I also took a quick look at the illegal exception thread. On my understanding, the error log happens due to a lack of participants during the initial process. Once there are enough participants to become alive, the controller will be able to rebalance automatically. The only difference between CrushEd and AutoRebalanceStrategy is that CrushEd complains about it and print an error log, while the AutoRebalanceStrategy silently returns an empty mapping without printing error log. Given this, it won't cause any trouble. Could you please elaborate on the valid scenario that you mentioned? If we are actually talking about different scenarios, we can help to revisit the logic and see if there is any issue in the CrushEd strategy.
---------------------------------------------------------------- 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] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
