[
https://issues.apache.org/jira/browse/HDFS-931?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12831784#action_12831784
]
Todd Lipcon commented on HDFS-931:
----------------------------------
Agreed. Do we dare just break them and put it in the release notes as an
incompatible change?
One reason for doing so is that close() already is listed as throwing
IOException. So it's not an API change, and in theory people *should* have
catch blocks in place.
> DFSClient.close doesn't rethrow exceptions thrown by automatically closed
> clients
> ---------------------------------------------------------------------------------
>
> Key: HDFS-931
> URL: https://issues.apache.org/jira/browse/HDFS-931
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs client
> Affects Versions: 0.21.0, 0.22.0
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Critical
> Attachments: hdfs-931.txt, hdfs-931.txt,
> TEST-org.apache.hadoop.hdfs.TestFiPipelines.txt
>
>
> I've written a fault injection test that simply throws exceptions on every
> ack.readFields call in ResponseProcessor. This ought to raise an exception to
> the client trying to use the stream, but instead it ignores it silently.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
