lhotari commented on pull request #9850: URL: https://github.com/apache/pulsar/pull/9850#issuecomment-796516578
> The solution I thought of: > I will start a new PR to add zk version to the policies cache. > When the policies cache in some thread is dirty(with old version), writing to zk will report an error. At this time, the latest cache will be read, and then this Function will be called again to ensure that the dirty cache will not overwrite the new data. Yes, I think that it's a good idea to handle the ZK and distributed updates issues separately. Please create a new issue for the data consistency issues around ZK and distributed modifications since that is a broader scope than #9711. My previous question was about preventing the "lost updates" problems that happen in the local data structures. I consider that problem a part of #9711. Please take a look if you could change the mutation logic to prevent the "lost updates" issue in the locations that were listed in my previous comment. @315157973 Could you resolve the local "lost updates" issues as part of this current PR, since after that I think it's ok to close 9711 with this PR? WDYT? ---------------------------------------------------------------- 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]
