[
https://issues.apache.org/jira/browse/IGNITE-14762?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aleksandr Polovtcev updated IGNITE-14762:
-----------------------------------------
Description:
{{RaftGroupServiceImpl#refreshLeader}} sometimes blocks when sending a request
from a node to itself.
h2. Probable cause
We suspect that {{DelegatingTransportFactory#requestResponseToSelf}} contains a
race condition: {{Flux}} returned by the {{listen()}} method might be lazily
evaluated, in which case the client will start listening for messages after the
message event handlers have already been executed, so no new incoming messages
will actually be detected.
was:{{RaftGroupServiceImpl#refreshLeader}} sometimes blocks when sending a
request from a node to itself.
> Leader refresh sometimes blocks indefinitely
> --------------------------------------------
>
> Key: IGNITE-14762
> URL: https://issues.apache.org/jira/browse/IGNITE-14762
> Project: Ignite
> Issue Type: Bug
> Reporter: Aleksandr Polovtcev
> Assignee: Aleksandr Polovtcev
> Priority: Blocker
> Labels: ignite-3, networking
> Fix For: 3.0.0-alpha2
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> {{RaftGroupServiceImpl#refreshLeader}} sometimes blocks when sending a
> request from a node to itself.
> h2. Probable cause
> We suspect that {{DelegatingTransportFactory#requestResponseToSelf}} contains
> a race condition: {{Flux}} returned by the {{listen()}} method might be
> lazily evaluated, in which case the client will start listening for messages
> after the message event handlers have already been executed, so no new
> incoming messages will actually be detected.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
