[
https://issues.apache.org/jira/browse/ACCUMULO-409?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13398974#comment-13398974
]
Keith Turner commented on ACCUMULO-409:
---------------------------------------
I found something interesting while looking into this. Tablet servers
currently copy failures and then the master copies them. This copy by the
tserver is not something that we could rely on to happen, it does not help fix
this problem in anyway. Its duplicate effort that needs to be removed. Its an
artifact of using the old bulk import code in FATE.
Also, its possible that a tablet server could copy something into the failed
dir, then a later import attempt by the master could succeed. This would
result in an unnecessary file in the failed dir.
> Failed bulk import files are copied through master
> --------------------------------------------------
>
> Key: ACCUMULO-409
> URL: https://issues.apache.org/jira/browse/ACCUMULO-409
> Project: Accumulo
> Issue Type: Bug
> Components: master
> Affects Versions: 1.4.0
> Reporter: Keith Turner
> Fix For: 1.4.2
>
>
> When bulk import fails, the failed files are copied into a fail dir by the
> master. This is very bad if its a large amount of data. Noticed this while
> debugging ACCUMULO-408.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
