impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Armstrong (Code Review)" <>
Subject [Impala-ASF-CR] IMPALA-5567: race in fragment instance teardown
Date Mon, 26 Jun 2017 21:01:09 GMT
Tim Armstrong has posted comments on this change.

Change subject: IMPALA-5567: race in fragment instance teardown

Patch Set 1:

File be/src/exec/

Line 101:     results_ = nullptr;
> At first glance this looks reasonable to put here, but is this necessary fo
I don't feel strongly. I think the idea was to move these status checks more closely to the
expr evaluations that could cause errors to be hit.
File be/src/runtime/runtime-state.h:

PS1, Line 294:   /// Returns a non-OK status if query execution should stop (e.g., the query
             :   /// cancelled or a mem limit was exceeded). Exec nodes should check this
periodically so
             :   /// execution doesn't continue if the query terminates abnormally. This should
not be
             :   /// called after ReleaseResources().
             :   Status CheckQueryState();
> I like CheckAsyncError() for GetQueryStatus(). If we can convert all of the
I think we should do the work of removing Allocate() calls at some point but I think we should
keep fix as targeted as possible. That might be an argument to defer doing the rename as well.

To view, visit
To unsubscribe, visit

Gerrit-MessageType: comment
Gerrit-Change-Id: If567e734042b5f2b82323368dd536dbf3bdf4744
Gerrit-PatchSet: 1
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Tim Armstrong <>
Gerrit-Reviewer: Impala Public Jenkins
Gerrit-Reviewer: Lars Volker <>
Gerrit-Reviewer: Matthew Jacobs <>
Gerrit-Reviewer: Michael Ho <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-HasComments: Yes

View raw message