[
https://issues.apache.org/jira/browse/HDFS-13778?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16609865#comment-16609865
]
Plamen Jeliazkov commented on HDFS-13778:
-----------------------------------------
Hey [~shv],
Couple things from me:
(1) Should we move this to the 'org.apache.hadoop.hdfs.server.namenode.ha'
package? Seems a more appropriate place.
(2) Regarding 'testMultiClientStatesWithRandomFailovers', what if we split the
load across two runs and only did 1 failover per run? The issue that I am
seeing on the Jenkins logs is clients being stuck waiting for each other. I
think we need to decrease number of clients as well. Optimal case is two
clients competing against each other. I will attach a patch (003) demonstrating
my idea.
(3) I attempted to resurrect 'testClientSendsState' but unfortunately I have no
way of accessing the ClientHAFactory needed to actually change the client's
AlignmentContext to some mocked one, nor does there seem to be anything else I
can mock to check if client is sending state as the AlignmentContext is the
only source for that.
> In TestStateAlignmentContextWithHA replace artificial
> AlignmentContextProxyProvider with real ObserverReadProxyProvider.
> ------------------------------------------------------------------------------------------------------------------------
>
> Key: HDFS-13778
> URL: https://issues.apache.org/jira/browse/HDFS-13778
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: test
> Reporter: Konstantin Shvachko
> Assignee: Sherwood Zheng
> Priority: Major
> Attachments: HDFS-13778-HDFS-12943.001.patch,
> HDFS-13778-HDFS-12943.002.patch
>
>
> TestStateAlignmentContextWithHA uses an artificial
> AlignmentContextProxyProvider, which was temporary needed for testing. Now
> that we have real ObserverReadProxyProvider it can take over ACPP. This is
> also useful for testing the ORPP.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
