[
https://issues.apache.org/jira/browse/HBASE-10181?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lars Hofhansl updated HBASE-10181:
----------------------------------
Attachment: 10181-alternate.txt
More "radical" change. Seems pointless to retry after a ClassNotFoundException.
If the de-serialization encountered an IOException, don't wrap again - this
obviously covers the DNRE scenario.
I'm fine with either changes. Would like to commit today, so that I can roll an
RC.
> HBaseObjectWritable.readObject catches DoNotRetryIOException and wraps it
> back in a regular IOException
> -------------------------------------------------------------------------------------------------------
>
> Key: HBASE-10181
> URL: https://issues.apache.org/jira/browse/HBASE-10181
> Project: HBase
> Issue Type: Bug
> Components: IPC/RPC
> Affects Versions: 0.94.12
> Reporter: Samarth
> Fix For: 0.94.15
>
> Attachments: 10181-alternate.txt, 10181.txt
>
>
> Exception handling inside HbaseObjectWritable needs to be reworked, IMHO.
> For example:
> At several places inside HbaseObjectWritable.readObject, exceptions are
> caught and rethrown as I/O Exception (including ClassNotFoundException!).
> So, if an implementation of readFields method throws a DoNotRetryIOException,
> HBase still ends up retrying.
> This problem exists at least in 0.94.12 version of HBase.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
