umamaheswararao commented on PR #3361: URL: https://github.com/apache/ozone/pull/3361#issuecomment-1122728552
Update: Today we had a small group discussion with the folks who involved in the review of above PR task. Attendees: @guihecheng , @sodonnel , @kaijchen , @adoroszlai , @umamaheswararao After discussing above options, we all came to the same agreement that, we will be going with Option #2, mainly because that could be quicker as much of needed APIs available in ContainerProtocolCalls. The disadvantages parts are also discussed When two co-ordinators doing the same work somehow, how do we prevent one from another. There could be an attempt ID when SCM scheduled, so that target nodes can check the attempt ID on writes to make sure to respect the right coordinator and reject others. Remaining solutions discussed in the PR comment already. For NW failures, we could still do local container writes and transfer if the current one really becoming an issues in realistic clusters. With the current approach and with improved retry mechanism, we could avoid NW glitches or simple restarts anyway. -- 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]
