You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@servicemix.apache.org by "Geert Schuring (JIRA)" <ji...@apache.org> on 2011/02/09 09:21:57 UTC
[jira] Issue Comment Edited: (SMX4NMR-195) Dynamically
enable/disable cluster engine when ConnectionFactory available in registry
[ https://issues.apache.org/jira/browse/SMX4NMR-195?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12992380#comment-12992380 ]
Geert Schuring edited comment on SMX4NMR-195 at 2/9/11 8:20 AM:
----------------------------------------------------------------
I agree. We should not ignore the fact that the NMR runs in an OSGi environment. If the connectionFactory leaves, so should the ClusterEngine. I'm not very happy with the blueprint container blocking until a service becomes available. We should set the timeout very short to make sure only very short periods between a service restart are covered. When a required service is gone, the cluster engine should receive an exception, or shutdown completely.
Otherwise we will see very hard to explain freezes. An OSGi application is supposed to 'fail early'.
was (Author: cathodion):
I agree. We should not ignore the fact that the NMR runs in an OSGi environment. If the connectionFactory leaves, so should the ClusterEngine. I'm not very happy with the blueprint container blocking until a service becomes available. We should set the timeout very short to make sure only very short periods between a service restart are covered. When a required service is gone, the cluster engine should receive an exception, or shutdown completely.
> Dynamically enable/disable cluster engine when ConnectionFactory available in registry
> --------------------------------------------------------------------------------------
>
> Key: SMX4NMR-195
> URL: https://issues.apache.org/jira/browse/SMX4NMR-195
> Project: ServiceMix NMR
> Issue Type: Improvement
> Affects Versions: 1.2.0
> Reporter: Gert Vanthienen
> Fix For: 1.4.0
>
>
> Currently, the Cluster engine is configured with a Blueprint proxy to the ConnectionFactory that's available in the OSGi Service Registry. This way, the Cluster Engine isn't aware of changes in the registry (e.g. the ConnectionFactory is being removed or a new one is registered).
> We should improve the engine to dynamically enable/disable clustering depending on the availability of a suitable ConnectionFactory.
--
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira