[
https://issues.apache.org/jira/browse/IMPALA-15285?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18107377#comment-18107377
]
Quanlong Huang commented on IMPALA-15285:
-----------------------------------------
Uploaded the query profile: [^agg_no_spilling_profile.txt]
Maybe after IMPALA-14594, the pre-agg tends to pass through rows instead of
expanding the hash table and spilling to disk. We can try decreasing
buffer_pool_limit like IMPALA-9452 did. Uploaded a patch:
https://gerrit.cloudera.org/c/24726/
> test_spilling_aggs fails with no spilling
> -----------------------------------------
>
> Key: IMPALA-15285
> URL: https://issues.apache.org/jira/browse/IMPALA-15285
> Project: IMPALA
> Issue Type: Bug
> Reporter: Quanlong Huang
> Assignee: Quanlong Huang
> Priority: Major
> Attachments: agg_no_spilling_profile.txt
>
>
> TestSpillingDebugActionDimensions.test_spilling_aggs fails intermittently due
> to the query is not spilling.
> {noformat}
> AssertionError: Aggregation of SUM over SpilledPartitions did not match
> expected results.
> EXPECTED VALUE:
> 0
>
>
> ACTUAL VALUE:
> 0
>
> OP:
> >{noformat}
> The query is the first one in
> testdata/workloads/functional-query/queries/QueryTest/spilling-aggs.test
> {code:sql}
> select l_orderkey, count(*)
> from lineitem
> group by 1
> order by 1 limit 10;{code}
> We might need to decrease buffer_pool_limit again like IMPALA-9452.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
