[
https://issues.apache.org/jira/browse/RATIS-2294?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tsz-wo Sze updated RATIS-2294:
------------------------------
Component/s: Netty
Summary: Fix NettyClientRpc for handling exception and timeout (was:
Fix TestRaftAsyncWithNetty Flaky test)
> Fix NettyClientRpc for handling exception and timeout
> -----------------------------------------------------
>
> Key: RATIS-2294
> URL: https://issues.apache.org/jira/browse/RATIS-2294
> Project: Ratis
> Issue Type: Improvement
> Components: Netty
> Reporter: Shilun Fan
> Assignee: Shilun Fan
> Priority: Major
> Attachments: 1264_review.patch
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> I'm currently follow up on RATIS-2251, and test results indicate that we have
> some flaky tests. I'm working on identifying and resolving the root causes.
> Specifically, I found that testStaleReadAsync and testStateMachineMetrics are
> failing because the client doesn't know the correct leaderId, which leads to
> request failures.
> In async mode, if the leaderId is not explicitly specified, the default used
> in tests is s0. However, during actual test execution with three peers (s0,
> s1, and s2), any of them can become the leader. Therefore, we need to
> explicitly set the correct leaderId for the client to ensure reliable
> communication.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
