[
https://issues.apache.org/jira/browse/IGNITE-11617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16800728#comment-16800728
]
Ivan Rakov commented on IGNITE-11617:
-------------------------------------
[~agoncharuk], fixed and merged to master.
> New exchange coordinator skips client fast reply for previous exchange
> ----------------------------------------------------------------------
>
> Key: IGNITE-11617
> URL: https://issues.apache.org/jira/browse/IGNITE-11617
> Project: Ignite
> Issue Type: Bug
> Reporter: Alexey Goncharuk
> Assignee: Ivan Rakov
> Priority: Critical
> Fix For: 2.8
>
> Attachments: ClientFastReplyCoordinatorFailureTest.java
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> The following scenario is broken:
> 1) A client joins topology, all server nodes complete client exchange, client
> delays sending it's single message
> 2) A new node joins topology, next exchange is started
> 3) Coordinator of the new exchange fails
> 4) New coordinator attempts to restore exchange state from all nodes,
> including client from step 1
> 5) Client send single message. The message should be processed by a
> fast-reply path because the corresponding exchange is completed. However,
> this does not happen because this exchange was completed with state {{SRV}}
> on new coordinator. Fast-reply is omitted and exchange hangs.
> Attached is a test reproducing the problem.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
