Henry Robinson has posted comments on this change. Change subject: IMPALA-4348 / IMPALA-4333: Improve coordinator fragment cancellation ......................................................................
Patch Set 2: (7 comments) http://gerrit.cloudera.org:8080/#/c/4865/2//COMMIT_MSG Commit Message: Line 13: access a destroyed PlanRootSink object and crash (IMPALA-4333). > did this bug reproduce in any tests, or was it only caught by Hue? Only caught by Hue. Needs some kind of... fault injection framework? That sounds like an idea. I tested it manually. Line 25: * Ensure that it is safe to call CloseConsumer() concurrently to > If I understand correctly there weren't any correctness fixes in this part, Done http://gerrit.cloudera.org:8080/#/c/4865/1/be/src/exec/plan-root-sink.cc File be/src/exec/plan-root-sink.cc: Line 163: > Yeah, I think we should be clearer what's necessary for correctness and wha Now that CloseConsumer() signals consumer_cv_ (which I think is an appropriate functional change), I think this is relevant to the patch (otherwise I agree, we should remove it). http://gerrit.cloudera.org:8080/#/c/4865/2/be/src/exec/plan-root-sink.h File be/src/exec/plan-root-sink.h: Line 79: /// that calls Send(). *eos is set to 'true' when there are no more rows to consume. > When does it block/unblock? Done http://gerrit.cloudera.org:8080/#/c/4865/2/be/src/runtime/coordinator.cc File be/src/runtime/coordinator.cc: PS2, Line 166: rpc_sent = true > We don't seem to do anything with this argument. Woops, thanks. It's ok for rpc_sent_ to be incorrectly set to true - it's not an error to try to cancel non-existant plan fragments - but it does prevent warnings in the logs. PS2, Line 517: might have been cancelled > what does it mean to "have been cancelled"? Since line 478 gets the lock, Failed is probably better, yep. Line 523: // Try and return the fragment instance status if it was already set. > Are there any situations where the fragment failed and this isn't set? Woul Reporting the failure happens asynchronously - I think it's very unlikely to be an issue, but there's nothing guaranteeing that the error is visible to this thread even though the fragment has finished. -- To view, visit http://gerrit.cloudera.org:8080/4865 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I5b4e25c1d658b3929182ba5e56b5c5e881dd394a Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Henry Robinson <[email protected]> Gerrit-Reviewer: Dan Hecht <[email protected]> Gerrit-Reviewer: Henry Robinson <[email protected]> Gerrit-Reviewer: Tim Armstrong <[email protected]> Gerrit-HasComments: Yes
