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-4458: Fix resource cleanup of cancelled mt scan nodes.
Date Wed, 30 Nov 2016 17:31:10 GMT
Tim Armstrong has posted comments on this change.

Change subject: IMPALA-4458: Fix resource cleanup of cancelled mt scan nodes.

Patch Set 2:


The change looks good aside from some minor comments. I think we should try running a local
stress test under ASAN with mt_dop set just to flush out remaining issues. I'm happy to do
that or help with that.
File be/src/exec/

PS2, Line 240: .get()
.get() is unnecessary

Line 251:   DCHECK_EQ(template_tuple_pool_.get()->total_allocated_bytes(), 0);
.get() is unnecessary here too

Line 278:   if (level_cache_pool_.get() != nullptr) {
Move this up to where we free the other pools for consistency
File be/src/exec/parquet-column-readers.h:

Line 232:   virtual void Close(RowBatch* row_batch) = 0;
Doc that row_batch can be NULL.

Line 346:     if (decompressed_data_pool_.get() != nullptr) {
This check looks unnecessary since it should never be non-NULL - it's initialised in the constructor
and then never reset.
File be/src/exec/scanner-context.h:

Line 258:     /// pool to batch.  If 'done' is set, releases the completed resources.
Mention that attaching only happens when both contains_tuple_data_ is true and done is false?

To view, visit
To unsubscribe, visit

Gerrit-MessageType: comment
Gerrit-Change-Id: Ib32f87b3289ed9e8fc2db0885675845e11207438
Gerrit-PatchSet: 2
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Alex Behm <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-HasComments: Yes

View raw message