[
https://issues.apache.org/jira/browse/DRILL-2650?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Krystal closed DRILL-2650.
--------------------------
git.commit.id.abbrev=0c1b293
Verified the durations of the individual fragments are within the duration of
the cancelled query itself.
> Cancelled queries json profile shows query end time occurs before fragments
> end time
> -------------------------------------------------------------------------------------
>
> Key: DRILL-2650
> URL: https://issues.apache.org/jira/browse/DRILL-2650
> Project: Apache Drill
> Issue Type: Bug
> Components: Client - HTTP
> Affects Versions: 0.9.0
> Reporter: Krystal
> Assignee: Sudheesh Katkam
> Fix For: 1.2.0
>
> Attachments: DRILL-2650.1.patch.txt
>
>
> I have a query that was successfully cancelled. The query start and end time
> is as follows:
> "type": 1,
> "start": 1427839192049,
> "end": 1427839194966,
> This translates to a query duration of about 3 seconds. However, the
> duration of the query's fragments are much longer up to more than 6 seconds.
> Here is an entry for majorFragmentId=0 with a duration of 6.6 seconds:
> "startTime": 1427839192796,
> "endTime": 1427839199408,
> 8 out of 11 major fragments have duration greater than the query itself. To
> an end user, this is confusing and does not make sense. We should wait for
> all of the major fragments to be completely cancelled before updating the the
> "end" time of the query.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
