narendly commented on issue #343: Fine-grained state transition throttling and respect MIN_ACTIVE replica URL: https://github.com/apache/helix/issues/343#issuecomment-512928795 The root cause for this problem is because we throttle things at partition granularity. For example, if Helix determines that a partition needs recovery balance, then it will subject all of its replicas to the recovery balance throttle config (although some replicas for the same partition might be undergoing load balance - these replicas will still be subject to recovery throttle, which is incorrect!). Enabling Helix to throttle at replica granularity will solve this problem. Hope this helps, Hunter Sent from BlueMail On Jul 18, 2019, 20:04, at 20:04, Yi Wang <notificati...@github.com> wrote: >I feel like we need to put more background knowledge and context on the >topic. For example, why Helix will treat the previous ongoing offline >-> slave as recovery balance given it was kicked off as load balance. >One naive question, I understand the current limitation is **helix >controller doesn't know the historical context about any ongoing state >transitions in a participant** (is it true). Is it because of the >**state transition message for the participant** get consumed? What if >we preserve the pending message until completion and controller would >be able to know the historical context? pros and cons? > >Let's put more background and context of the problem to have a >meaningful discussion. > >-- >You are receiving this because you are subscribed to this thread. >Reply to this email directly or view it on GitHub: >https://github.com/apache/helix/issues/343#issuecomment-512924648
---------------------------------------------------------------- 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 With regards, Apache Git Services
