Marcel Kornacker has posted comments on this change. Change subject: IMPALA-4890/5143: Coordinator race involving TearDown() ......................................................................
Patch Set 1: (1 comment) http://gerrit.cloudera.org:8080/#/c/6897/1/be/src/runtime/coordinator.cc File be/src/runtime/coordinator.cc: Line 895 > This TODO still seems relevant - WaitForBackendCompletion() shouldn't happe i think the waitforbackendcompletion call should be removed altogether, but maybe i'm missing something: - we're releasing resources automatically once the result set has been returned - there is no more teardown - post-query finalization only applies to insert statements, and those don't end up here however, we should only release resources once the last result rows have been returned/processed, ie, planrootsink::getnext returning rows in 'results' and setting eos to true simultaneously would be problematic, i think. i didn't quite follow the eos logic in planrootsink. does it set eos=true only when it returns no rows? -- To view, visit http://gerrit.cloudera.org:8080/6897 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I457a6424a0255c137336c4bc01a6e7ed830d18c7 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Marcel Kornacker <mar...@cloudera.com> Gerrit-Reviewer: Henry Robinson <he...@cloudera.com> Gerrit-Reviewer: Marcel Kornacker <mar...@cloudera.com> Gerrit-HasComments: Yes