[
https://issues.apache.org/jira/browse/CHUKWA-75?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12694786#action_12694786
]
Jerome Boulon commented on CHUKWA-75:
-------------------------------------
If we don't know the column that we are processing how can we be sure that a
value >1000000000L is an overflow?
- Should we or not apply the overflow only on a specific list of headers that
we are aware of?
Also, can we have a validation that will not create any exception from Demux,
DumpRecord, pig?
Having a null there will force everyone than want to use this recordType to
check for NULL (performance impact + check for null at query time),
so in the worse case it should be documented but having a well defined value
in place of NULL will be better.
> iostat output overflow
> ----------------------
>
> Key: CHUKWA-75
> URL: https://issues.apache.org/jira/browse/CHUKWA-75
> Project: Hadoop Chukwa
> Issue Type: Bug
> Components: Data Processors
> Environment: Redhat 5.1, Java 6
> Reporter: Eric Yang
> Assignee: Eric Yang
> Attachments: CHUKWA-75-1.patch, CHUKWA-75.patch
>
>
> On older systems, iostat metrics contain random overflow. The plan is the
> put in a filter to remove the overflow in the demux parser to improve the
> data quality.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
