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-1575: Part 1: eagerly release query exec resources
Date Fri, 27 Oct 2017 20:57:53 GMT
Tim Armstrong has posted comments on this change. ( )

Change subject: IMPALA-1575: Part 1: eagerly release query exec resources

Patch Set 11:

File be/src/runtime/
PS11, Line 359: was not available
> Okay, but my original question was whether the comment should be reworded f
File be/src/runtime/
PS11, Line 95:   if (!released_exec_resources_) ReleaseExecResources();
> What I proposed was not to handle it in Init(), but let the caller handle i
Should have read it more carefully, sorry. I try to avoid function APIs that require callers
to undo side-effects of a failed function - that seems error-prone in general but I guess
it is ok in this specific case.
File be/src/runtime/
PS11, Line 88:   local_query_state_->AcquireExecResourceRefcount(); // Decremented in ReleaseResources().
> Why are exvaluating exprs query-wide resources. Aren't they per-thread reso
I'd consider per-thread execution resources to be a subset of query-wide execution resources.
I.e. releasing a query's execution resources while threads are still consuming execution resources
is not valid in my mind. The comments in QueryState weren't very clear about what holding
a refcount meant so I tried to improve them.

To view, visit
To unsubscribe, visit

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I41ff374b0403f10a145f7fee9b3145953ee32341
Gerrit-Change-Number: 8303
Gerrit-PatchSet: 11
Gerrit-Owner: Tim Armstrong <>
Gerrit-Reviewer: Dan Hecht <>
Gerrit-Reviewer: Joe McDonnell <>
Gerrit-Reviewer: Sailesh Mukil <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-Comment-Date: Fri, 27 Oct 2017 20:57:53 +0000
Gerrit-HasComments: Yes

  • Unnamed multipart/alternative (inline, 8-Bit, 0 bytes)
View raw message