ayushtkn commented on PR #6244: URL: https://github.com/apache/hadoop/pull/6244#issuecomment-1811994921
I have been reading the comments on the original ticket, and mostly share same thought as @Hexiaoqiao, the one case in the end mentioned was answered here, I believe: https://issues.apache.org/jira/browse/HDFS-16016?focusedCommentId=17783162&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-17783162 and the reporter I believe tried without [HDFS-16898](https://issues.apache.org/jira/browse/HDFS-16898)? In general I didn't see IBR/FBR chocking at Datanode side, FBR anyway is very low frequency, the only place FBR creates problem is at NN side, when almost all DN shoot at one time, there is a logic to handle this scenario as well, Anyway..... I was under the impression like when a DN pushes a FBR, all the pending IBR are pushed before? it isn't that the case now? What is the fix, put both IBR & BR under lock? earlier only IBR was under lock or what? what case was the earlier lock solving? I am just curious, I am not gonna merge this in general, will let other folks involved in the original ticket chase this.. -- 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]
