[ 
https://issues.apache.org/jira/browse/IMPALA-13411?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17890754#comment-17890754
 ] 

ASF subversion and git services commented on IMPALA-13411:
----------------------------------------------------------

Commit f5a8771a982cb270bbd7bce8e8253198c35c5ab4 in impala's branch 
refs/heads/master from Yida Wu
[ https://gitbox.apache.org/repos/asf?p=impala.git;h=f5a8771a9 ]

IMPALA-13411: Fix DCHECK fires for scan nodes that produce zero-length tuples

Removed the DCHECK assertion that tuple_data_len must be greater
than zero in tuple-file-writer.cc and tuple-file-reader.cc,
because in certain cases, such as count(*), tuple_data_len can
be zero, as no column data is returned and only the row count
matters.

Tests:
Adds TestTupleCacheCountStar for the regression test.

Change-Id: I264b537f0eb678b65081e90c21726198f254513d
Reviewed-on: http://gerrit.cloudera.org:8080/21953
Reviewed-by: Impala Public Jenkins <[email protected]>
Tested-by: Impala Public Jenkins <[email protected]>


> DCHECK can fire for scan nodes that produce zero-length tuples
> --------------------------------------------------------------
>
>                 Key: IMPALA-13411
>                 URL: https://issues.apache.org/jira/browse/IMPALA-13411
>             Project: IMPALA
>          Issue Type: Bug
>          Components: Backend
>    Affects Versions: Impala 4.5.0
>            Reporter: Joe McDonnell
>            Assignee: Michael Smith
>            Priority: Major
>
> This query crashes when using the tuple cache:
> {noformat}
> select count(*) from functional.alltypes;{noformat}
> It hits this assert: 
> [https://github.com/apache/impala/blob/6121c4f7d61fb9f2341cf14e1be3404325fb35b9/be/src/exec/tuple-file-writer.cc#L105]
> {noformat}
>   DCHECK_GT(tuple_data_len, 0);{noformat}
> For this count(*) query, it is not actually returning any columns. Instead, 
> only the row count matters and the count does not require any tuple data. So, 
> the tuple data is empty.
> We can revisit this DCHECK and either remove it or modify the condition to 
> check if the tuple size is zero.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to