[jboss-jira] [JBoss JIRA] Commented: (AS7-1751) infinispan setup to support environment variables within the standalone.xml file
Paul Ferraro (JIRA)
jira-events at lists.jboss.org
Fri Sep 9 10:54:26 EDT 2011
[ https://issues.jboss.org/browse/AS7-1751?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12627560#comment-12627560 ]
Paul Ferraro commented on AS7-1751:
-----------------------------------
The recommended way to isolate clusters (i.e. oam vs traffic) in AS7 is to use either a different multicast address, or different multicast port. By default these are manipulated via the "jgroups-udp" socket binding.
This is conceptually different from multiple jgroups channels on the same server, since these channel will use the same shared transport layer - and can be thought of as a single "cluster".
But your point is still valid - you should have the option to manipulate the cluster name independently from the cache-container name. Something like:
<cache-container name="web" ...>
<transport cluster="${CLUSTER}-web"/>
</cache-container>
Alternatively (since it might be annoying to have to do this within every cache container) or perhaps additionally, we add this logic to the jgroups subsystem itself. Something like:
<subsystem xmlns="urn:jboss:domain:jgroups:1.0" default-stack="udp" cluster-prefix="${CLUSTER}-">
...
</subsystem>
Thoughts?
> infinispan setup to support environment variables within the standalone.xml file
> --------------------------------------------------------------------------------
>
> Key: AS7-1751
> URL: https://issues.jboss.org/browse/AS7-1751
> Project: Application Server 7
> Issue Type: Enhancement
> Components: Clustering
> Affects Versions: 7.0.1.Final
> Reporter: Dan Sirbu
> Assignee: Paul Ferraro
> Fix For: 7.0.2.Final
>
>
> When I kickstart the inifispan subsystem by using the start=EAGER option, saw that there are multiple "clusters" that get kickstarted as:
> GMS: address=traffic_instance_PL-3-31315, cluster=hibernate
> ........................................, cluster=web
> and others.
> Now, if I do have mutliple blades e.g. SC-1, SC-2, PL-3, PL-4 and then I would like to define two clusters e.g. "oam" , "traffic" where "oam" should be assigned for SC-x and "traffic" for PL-x, then I will certainly need to go in the standalone-ha.xml file and change all the "cache-container name" entries in order to have "oam" cluster distinguished from "traffic" cluster, right ?
> If so, it would be better if for example I could pass an jvm option via e.g. -DCLUSTER="oam" and then in the standalone-ha.xml file have this set as: "cache-container name=${CLUSTER}-"
> In this way, I can obtain a dynamic config at runtime.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
More information about the jboss-jira
mailing list