[GitHub] [hadoop] ayushtkn commented on pull request #5445: HDFS-16938. Utility to trigger heartbeat and wait until BP thread queue is fully processed
ayushtkn commented on PR #5445: URL: https://github.com/apache/hadoop/pull/5445#issuecomment-1451100991 Playing a bit more with it: Single sleep doesn't repro itself. My wild guess is, it isn't testing from where we started. It is inducing a race by **nextHeartbeatTime**, that too because we induced sleeps, else like very tough with the kind of code we have now. > we might rather want to wait for source code to do that so that if something changes in source code sequence or so, our test would be able to catch it not the intention of the original test, it is just waiting for the message in the Queue to be sent and processed and checking the response of namenode to that, or if namenode acknowledges that or not. (I still feel we could have invoked **processQueueMessages** directly and saved some time, but lets see, the present code is also working) If someone removes the processQueueMessages itself from the **sendHeartbeat**, then also this test should fail or atleast some should, rest we can't guarantee everything... >but anyways nothing wrong with waiting for sometime and circling back to this when/if required. yeps, lets wait and see if any failures and circle back in future and see what are the potential solutions then. -- 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: common-issues-unsubscr...@hadoop.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[GitHub] [hadoop] ayushtkn commented on pull request #5445: HDFS-16938. Utility to trigger heartbeat and wait until BP thread queue is fully processed
ayushtkn commented on PR #5445: URL: https://github.com/apache/hadoop/pull/5445#issuecomment-1451064227 First patch isn't something to consider itself. Second one is like very unrealistic in general, there aren't any things which we can blame above which can consume time, they are some naive if checks or so, can't take this time... I don't think we have a use case for now for this util, If my intent is just for processQueueMessages, I will expose and just shoot that directly, rather than doing the whole loop. Changing the existing uses with this would be slowing down tests, which are already above tolerable limits. Lets hold it, and observe if we get something around this in future we can circle back -- 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: common-issues-unsubscr...@hadoop.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
[GitHub] [hadoop] ayushtkn commented on pull request #5445: HDFS-16938. Utility to trigger heartbeat and wait until BP thread queue is fully processed
ayushtkn commented on PR #5445: URL: https://github.com/apache/hadoop/pull/5445#issuecomment-1450903655 Viraj, can you help me repro the scenario. I added 10K sleep before processEnqueed method last time and the test didn’t fail for me, triggerHeartbeat was working there -- 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: common-issues-unsubscr...@hadoop.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org - To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-issues-h...@hadoop.apache.org
