]
Radim Vansa reopened ISPN-3942:
-------------------------------
The fix does nothing. We cannot rely on another thread miraculously updating the server
list - we have to check that the provided address was not already tried.
Making the field volatile is unnecessary as the balancer is called only in synchronized
fashion.
HotRod client keep trying recover a connection to a cluster member
after shutdown
---------------------------------------------------------------------------------
Key: ISPN-3942
URL:
https://issues.jboss.org/browse/ISPN-3942
Project: Infinispan
Issue Type: Bug
Components: Remote Protocols
Affects Versions: 6.0.1.Final, 7.0.0.Alpha1
Environment: Infinispan server with HotRod client in server-client mode and
replicated caches.
Reporter: Wolf-Dieter Fink
Assignee: Pedro Ruivo
Priority: Critical
Labels: clustered, hotrod, hotrod-java-client
Fix For: 7.0.0.Alpha1, 7.0.0.Final
Attachments: hotrod-client.log
If a HotRod client is connected to a server cluster and one of the servers do a correct
shutdown, the client keep try to reconnect the server after the cluster view has changed.
There is only a replicated cache configured.
From the server-logfile the new cluster view is established
16:00:22,290 INFO [org.infinispan.remoting.transport.jgroups.JGroupsTransport]
(Incoming-3,shared=udp) ISPN000094: Received new cluster view: [node1/clustered|2] (1)
[node1/clustered]
There is no failure from the application perspective, but there there is always a retry
if the RoundRobin return the unavailable server.
This might end in a performance or failure issue if there is a larger cluster and a part
going out of work for some reason.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: