[
http://jira.jboss.com/jira/browse/JBMESSAGING-761?page=all ]
Tim Fox updated JBMESSAGING-761:
--------------------------------
We should probably introduce a "no server failover" mode, which if true, means
no server side failover takes place.
This can be used in conjunction with supportsFailover = false on the connection factory
(actually we should force supportsFailover to false if no server failover = true).
This means if a node fails or is taken out of the cluster for maintenance, then the client
will receive an exception and can retry looking up the CF using HAJNDI and reconnecting
manually.
In other words similar to JBoss MQ, except we would still have client side load balancing,
so we can exploit the cluster.
Consider clean ways of taking a node out of the cluster for
maintenance
-----------------------------------------------------------------------
Key: JBMESSAGING-761
URL:
http://jira.jboss.com/jira/browse/JBMESSAGING-761
Project: JBoss Messaging
Issue Type: Task
Reporter: Tim Fox
Fix For: 1.4.0.CR1
We should consider how the system administrator can take a node out of the cluster for
maintenance without clients losing messages.
If a node is just shut down then any clients connected to that node will failover to
another node, potentially losing any non persistent messages - this may not be
acceptable.
We should also consider functionality on the jmx console where a server cen block the
creation of any new connections, and the information such as client machine name, client
id, ip address can be displayed for existing connetions. This would allow the sysadmin to
contact the clients and shut them down cleanly before bringing the server down.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira