[
https://issues.apache.org/jira/browse/HBASE-8318?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13632269#comment-13632269
]
Jean-Daniel Cryans commented on HBASE-8318:
-------------------------------------------
bq. In total, I have about 400 tasks. So if 1 is failling, I can expect a
maximum of 0.25% errors. Which is totally accceptable (again, for this specific
usecase).
FWIW if you have a machine failure then you'll usually retry as many maps as
map slots are available on your TT.
bq. We have almost the same issue with the Puts where we will loose some of the
versions if we have to re-launch some tasks.
Writing from MR should always be idempotent, because you don't know exactly how
many times you'll write each row. If a task fails midway, then the rows already
written will be later overwritten and that's fine. So I don't see how we have
"almost the same issue".
bq. So as long as the user is aware of the risks I think this JIRA is
acceptable.
How would that be communicated to the user? It seems to me that just leaving a
comment in the javadoc is insufficient considering that speculative exec is
turned on by default, meaning that double/triple counting would not just be an
edge case but the default case.
> TableOutputFormat.TableRecordWriter should accept Increments
> ------------------------------------------------------------
>
> Key: HBASE-8318
> URL: https://issues.apache.org/jira/browse/HBASE-8318
> Project: HBase
> Issue Type: Bug
> Reporter: Jean-Marc Spaggiari
> Assignee: Jean-Marc Spaggiari
> Attachments: HBASE-8318-v0-trunk.patch, HBASE-8318-v1-trunk.patch,
> HBASE-8318-v2-trunk.patch
>
>
> TableOutputFormat.TableRecordWriter can take Puts and Deletes but it should
> also accept Increments.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
