[
https://issues.apache.org/jira/browse/PHOENIX-1429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14208564#comment-14208564
]
Eli Levine commented on PHOENIX-1429:
-------------------------------------
Looking over the patch...
In MergeSortResultIterator you left the
{code}
if (iterators != null) {
SQLCloseables.closeAll(iterators);
}
{code}
at the bottom of the *close()* method. Intentional? The same code is invoked in
the *finally* block in the new code.
> Cancel queued threads when limit reached
> ----------------------------------------
>
> Key: PHOENIX-1429
> URL: https://issues.apache.org/jira/browse/PHOENIX-1429
> Project: Phoenix
> Issue Type: Bug
> Reporter: James Taylor
> Attachments: PHOENIX-1429.patch
>
>
> We currently spawn a thread per guidepost to process a query. For a full scan
> with a LIMIT of 1, this is particularly inefficient, as once we get our one
> row back, we no longer need to execute the other queued threads. We should
> cancel those threads once the limit is reached.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
