[
https://issues.apache.org/jira/browse/HADOOP-6679?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12853925#action_12853925
]
Steve Loughran commented on HADOOP-6679:
----------------------------------------
# This is a git patch, not something hudson will handle, try --no-prefix, I
think.
# What JVM options are you running to disable DNS caching, both successful and
unsuccessful. Without that, retry in code may not be useful.
# If we are changing how ports are opened, I'd quite like to see a string
parameter saying what the port is to be used for, perhaps with some logging.
Sometimes it can be tricky trying to find out why some apparenlty random port
is open, and if it was all logged (at -debug, with a stack trace), some things
may be easier to deal with.
> add retry logic to NetUtils.createSocketAddr() when hostname resolution fails
> -----------------------------------------------------------------------------
>
> Key: HADOOP-6679
> URL: https://issues.apache.org/jira/browse/HADOOP-6679
> Project: Hadoop Common
> Issue Type: Improvement
> Affects Versions: 0.20.0
> Reporter: sam rash
> Priority: Minor
> Attachments: HADOOP-6679-patch-1.txt
>
>
> transient DNS errors can cause the InetSocketAddress created in
> NetUtils.createSocketAddr() to be unresolved when the hostname is resolvable.
> Add a single retry after a short sleep using a simple
> InetSocketAddressFactory class
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
