[jira] [Commented] (RATIS-844) RaftClient should connect to the new leader when the old leader is partitioned from the cluster

[Lokesh Jain (Jira)]

    [ 
https://issues.apache.org/jira/browse/RATIS-844?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17090559#comment-17090559
 ] 

Lokesh Jain commented on RATIS-844:
-----------------------------------

[~burcukozkan] Thanks for updating the patch! The changes look good to me. 
Please find my comments below.
 # The added test fails for Netty.
 # We will need to enable test exclusively for Grpc and Netty in 
TestLeaderElectionWithGrpc and TestLeaderElectionWithNetty respectively. We can 
remove the @Test annotation in LeaderElectionTests for the function and call 
the function in the respective test classes.
 # Unused import in NettyRpcService.

> RaftClient should connect to the new leader when the old leader is 
> partitioned from the cluster
> -----------------------------------------------------------------------------------------------
>
>                 Key: RATIS-844
>                 URL: https://issues.apache.org/jira/browse/RATIS-844
>             Project: Ratis
>          Issue Type: Bug
>          Components: client
>    Affects Versions: 0.6.0
>            Reporter: Burcu Ozkan
>            Priority: Major
>         Attachments: RATIS-844.001.patch, RATIS-844.002.patch
>
>
> While running a Netty cluster, RaftClient is blocked after it submits a 
> request to the leader which is partitioned from the cluster.
> The attached test reproduces the problem. In the test, after the client 
> connects to the leader, the leader is disconnected and another node is 
> elected as the new leader. However, the client waits indefinitely for a reply 
> from the previous one.
> The test passes for Grpc and HadoopRpc, but fails with TestTimedOutException 
> for Netty.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)