Abhishek Rawat has posted comments on this change. ( http://gerrit.cloudera.org:8080/22422 )
Change subject: IMPALA-13703: Cancel running queries before shutdown deadline ...................................................................... Patch Set 7: (1 comment) http://gerrit.cloudera.org:8080/#/c/22422/7/be/src/service/cancellation-work.h File be/src/service/cancellation-work.h: http://gerrit.cloudera.org:8080/#/c/22422/7/be/src/service/cancellation-work.h@32 PS7, Line 32: // The query is being terminated because a backend failed. We can skip cancelling the : // query if the fragment instances running on that backend all completed. : BACKEND_FAILED, : // The Impala Server terminated the query during shutdown and the query has not been : // done yet when the graceful shutdown deadline is reached. : GRACEFUL_SHUTDOWN Could we have a conflict between these two cancellation flow? In some cases executor daemons could be shutdown first and once they've shutdown coordinator daemons are shutdown. It's possible that the coordinator may start cancelling the queries also because it thinks backend has failed/cancelled? Can we add a test case for that scenario where we shutdown executors first with queries in flight and we shutdown coordinators at the end? -- To view, visit http://gerrit.cloudera.org:8080/22422 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I1cac2e100d329644e21fdceb0b23901b08079130 Gerrit-Change-Number: 22422 Gerrit-PatchSet: 7 Gerrit-Owner: Yida Wu <[email protected]> Gerrit-Reviewer: Abhishek Rawat <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Yida Wu <[email protected]> Gerrit-Comment-Date: Tue, 04 Feb 2025 20:30:46 +0000 Gerrit-HasComments: Yes
