[
https://issues.apache.org/jira/browse/HADOOP-17451?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17259724#comment-17259724
]
Steve Loughran commented on HADOOP-17451:
-----------------------------------------
..doesn't reoccur: network issues. For all but one they surface as network or
AWS client IOEs. The RandomIO test recovers from the failures but the stats
report
(stream_read_exceptions=7)
(stream_aborted=7)
(stream_read_opened=11)
the assertion was that stream_read_opened == 4. Strictly it should be
(stream_read_opened - stream_read_exceptions) ==4. To fix it I'd need to extend
the StreamStatistics interface to report the exception count, then change the
assertion. But since it will only surface on a network error, where lots of
other things fail: WONTFIX.
> intermittent failure of S3A huge file upload tests: count of bytes uploaded
> == 0
> --------------------------------------------------------------------------------
>
> Key: HADOOP-17451
> URL: https://issues.apache.org/jira/browse/HADOOP-17451
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 3.4.0
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Priority: Major
> Labels: pull-request-available
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> Intermittent failure of ITestHuge* upload tests, when doing parallel test
> runs.
> The count of bytes uploaded through StorageStatistics isn't updated. Maybe
> the expected counter isn't updated, and somehow in a parallel run with
> recycled FS instances/set up directory structure this surfaces the way it
> doesn't in a single test run.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
