[
https://issues.apache.org/jira/browse/ARTEMIS-3742?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tobias Månsson updated ARTEMIS-3742:
------------------------------------
Description:
We have issues were multiple connections are setup to the same cluster node
after a uncontrolled node failure. This resulted in a different node having two
connections setup to that failed one.
This is reproducible using docker by setting up a static cluster connection
with two nodes (artemis-0 & artemis-1) and running "docker rm -f artemis-1".
After restarting artemis-1, artemis-0 has two cluster connection to artemis-1.
This caused issues with queue binding in the form of AMQ224037 errors.
Is this expected to occur? Can cluster connections be removed trough the
management api? I only found that can be stopped.
was:
We have a process were we use the same hostname for cluster nodes as part of
node upgrades.
This is not ideal, and we are working to change it.
We've had issues now though were multiple connections are setup to the same
cluster node, one per physical host with the same hostname. This is probably to
be expected, but what was not expected was that the connections persisted after
the old node was taken down. This resulted in a different node having two
connections setup, but both to the same physical node.
This caused issues with queue binding in the form of AMQ224037 errors.
Is this expected to occur? Can cluster connections be removed trough the
management api? I only found that can be stopped.
> Multiple cluster connections are made to the same cluster node
> --------------------------------------------------------------
>
> Key: ARTEMIS-3742
> URL: https://issues.apache.org/jira/browse/ARTEMIS-3742
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Reporter: Tobias Månsson
> Priority: Major
>
> We have issues were multiple connections are setup to the same cluster node
> after a uncontrolled node failure. This resulted in a different node having
> two connections setup to that failed one.
> This is reproducible using docker by setting up a static cluster connection
> with two nodes (artemis-0 & artemis-1) and running "docker rm -f artemis-1".
> After restarting artemis-1, artemis-0 has two cluster connection to artemis-1.
>
> This caused issues with queue binding in the form of AMQ224037 errors.
> Is this expected to occur? Can cluster connections be removed trough the
> management api? I only found that can be stopped.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
