Riza Suminto has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/18138 )

Change subject: IMPALA-11072: Deflake 
TestSpillingDebugActionDimensions.test_spilling
......................................................................


Patch Set 1:

> Patch Set 1: Code-Review+1
>
> Seems good to me. Have you run the test several times so that we can 
> reasonably conclude that the flakiness is fixed? Or is it not feasible?

Yes, I have ran it few times and both hash join spill consistently.

Before I lower buffer_pool_limit, the join builder id=4 only spill 2 
partitions, which I think is very feasible to not spill at all if the fragment 
instances does not overlap long enough.

This tests used to run with just 15MB buffer_pool_limit. Commit fb5dc9eb48 
raise it to 215MB, but is probably too big to ensure consistent spilling.


--
To view, visit http://gerrit.cloudera.org:8080/18138
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Ie8802505e0dcae1be5e855107436805bd10e0077
Gerrit-Change-Number: 18138
Gerrit-PatchSet: 1
Gerrit-Owner: Riza Suminto <[email protected]>
Gerrit-Reviewer: Daniel Becker <[email protected]>
Gerrit-Reviewer: Impala Public Jenkins <[email protected]>
Gerrit-Reviewer: Riza Suminto <[email protected]>
Gerrit-Comment-Date: Mon, 10 Jan 2022 22:58:19 +0000
Gerrit-HasComments: No

Reply via email to