[
https://issues.apache.org/jira/browse/HDFS-9797?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15144027#comment-15144027
]
Inigo Goiri commented on HDFS-9797:
-----------------------------------
With HDFS-6440, now we can have more than two NNs. With the current
{{RequestHedgingProxyProvider}}, if we have 4 NNs, we see:
Invocation returned exception on [hostA]
Invocation returned exception on [hostC]
Invocation returned exception on [hostD]
Those exception are usually caused by trying to query the Standby NN. Given
that this is the expected behavior, I propose to check the exception and just
show a debugging message for those.
> RequestHedgingProxyProvider is too verbose with Standby exceptions
> ------------------------------------------------------------------
>
> Key: HDFS-9797
> URL: https://issues.apache.org/jira/browse/HDFS-9797
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs-client
> Affects Versions: 2.8.0
> Reporter: Inigo Goiri
> Assignee: Inigo Goiri
> Priority: Minor
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> {{RequestHedgingProxyProvider}} tries to connect to all the Namenodes and
> reports the exception for all the standby exceptions for all the other
> namenodes. There is no point on reporting the standby exception if it's
> expected.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
