[
https://issues.apache.org/jira/browse/HDFS-13779?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16598050#comment-16598050
]
Erik Krogen commented on HDFS-13779:
------------------------------------
Looks like me and Konstantin hit a race condition :) Comments 2 and 4 were
fixed in v001 patch.
I just attached v002 which addresses 1 and 3. Konstantin and I discussed
offline and realized that the caching done in {{ipc.Client}} will make the two
ProxyProviders share a single connection per NameNode regardless of the caching
of the proxies themselves, so I agree that the {{CachingHAProxyFactory}} is not
necessary. This also makes my changes to the constructor (accepting a class and
using reflection, instead of directly accepting an instance) unnecessary.
> Implement performFailover logic for ObserverReadProxyProvider.
> --------------------------------------------------------------
>
> Key: HDFS-13779
> URL: https://issues.apache.org/jira/browse/HDFS-13779
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: hdfs-client
> Reporter: Konstantin Shvachko
> Assignee: Erik Krogen
> Priority: Major
> Attachments: HDFS-13779-HDFS-12943.000.patch,
> HDFS-13779-HDFS-12943.001.patch, HDFS-13779-HDFS-12943.002.patch,
> HDFS-13779-HDFS-12943.WIP00.patch
>
>
> Currently {{ObserverReadProxyProvider}} inherits {{performFailover()}} method
> from {{ConfiguredFailoverProxyProvider}}, which simply increments the index
> and switches over to another NameNode. The logic for ORPP should be smart
> enough to choose another observer, otherwise it can switch to a SBN, where
> reads are disallowed, or to an ANN, which defeats the purpose of reads from
> standby.
> This was discussed in HDFS-12976.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
