[
https://issues.apache.org/jira/browse/CHUKWA-506?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
William Bajzek updated CHUKWA-506:
----------------------------------
Attachment: postProcessorRetry.patch
A patch which implements this. I have tested it out by starting chukwa,
shutting down MySQL, and then submitting a job. after the SQLException appeared
in the postprocessor.log, I restarted MySQL and waited. The next time the
postprocessor ran, the data was correctly inserted into the database.
Unfortunately, I can't get the unit tests to pass even with an unchanged,
freshly checked out copy of chukwa. They worked for me the other day, but not
now, but I was told on chukwa-dev to submit the patch for review anyway.
> Add retry logic to MetricDataLoader
> -----------------------------------
>
> Key: CHUKWA-506
> URL: https://issues.apache.org/jira/browse/CHUKWA-506
> Project: Chukwa
> Issue Type: Improvement
> Affects Versions: 0.4.0
> Reporter: William Bajzek
> Priority: Minor
> Attachments: postProcessorRetry.patch
>
>
> When processing data in Chukwa, many steps along the way, when the encounter
> an error condition, will attempt to reprocess the data a few times before
> giving up. This doesn't happen in post processing (at least when using
> MetricDataLoader); for example if there is a SQLException of any kind, it
> will move the data to the "inError" directory at which point it becomes AFAIK
> a manual process to move the data back to be reprocessed, so it would be nice
> to retry a few times in case the exception was transient (database was being
> bounced, or similar).
> I will attach a patch to be reviewed.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
