[
https://jira.jboss.org/jira/browse/JBAS-7254?page=com.atlassian.jira.plug...
]
Brian Stansberry commented on JBAS-7254:
----------------------------------------
Assigned to myself. I've been doing a lot of work on ClusterPartition these days and
at the same time we've been discussing the JGroups UUID.get() method as the source for
the values passed to Channel.setName(). Since I'm messing with this code, it's
easy enough for me to do.
Note that contrary to the description, I'm not going to deprecate verifyNodeIsUnique.
JGroups logical addresses resolve the problems JGroups has internally with node
reincarnation, but they don't resolve the problems an application has (e.g. no
response to RPC from the zombie member.)
Remove ChannelFactory additional_data usage
-------------------------------------------
Key: JBAS-7254
URL:
https://jira.jboss.org/jira/browse/JBAS-7254
Project: JBoss Application Server
Issue Type: Task
Security Level: Public(Everyone can see)
Components: Clustering
Reporter: Brian Stansberry
Assignee: Brian Stansberry
Fix For: 6.0.0.CR1
org.jboss.ha.framework.server.JChannelFactory supports a form of logical addresses by
setting additional_data on the channel (see setChannelUniqueId()). This should be
replaced by calling the JGroups 2.8 Channel.setName() method.
The ClusterPartition.verifyNodeIsUnique() method uses the additional_data to try to
ensure that the same node cannot rejoin the group until its previous incarnation has been
removed from the view. Confirm that JGroups 2.8 handles this itself, removing the need for
the AS to do it. Then make verifyNodeIsUnique a deprecated no-op (if this JIRA is done for
AS 5.2) and remove it for AS 6. Can't just remove the method in 5.2 as it is
protected; some subclass somewhere may invoke or override it.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira