You bring up a good point. I started a thread the other day about replicated datastores for service management, however no one chimed in. In the event that you use a single LDAP or database server, then this problem is simple as you merely need to run the Quartz job to reload the other nodes.
However, this causes you to have a single point of failure. Normally, you would handle this by load balancing between the LDAP or database servers, however this now requires that you replicate between the LDAP or database servers. What are your thoughts about handling service management in a HA environment (2+ CAS servers)? On 7/15/09 3:02 PM, "Marvin Addison" <[email protected]> wrote: >> I would have assumed that the default services >> manager call the reload after a service is added, updated, or deleted. > > I don't think that will fix the problem. The node that made the > change is self-consistent. It's the _other_ CAS nodes that have stale > data and need to call reload(). > >> I will file a JIRA issue for this as this is something that should be >> addressed in some form or fashion. > > Thanks for taking the initiative to file a Jira issue for this; it > clearly needs to be addressed. I can't think of any straightforward > solution other than a Quartz job that calls reload() on the > ServicesManager bean periodically. That way the other nodes will > eventually refresh and get updated. Ideally there'd be some sort of > message-driven notification to other nodes, but that seems a lot more > complex. > > M -- Andrew Feller, Business System Programmer LSU University Information Services 200 Frey Computing Services Center Baton Rouge, LA 70803 Office: 225.578.3737 Fax: 225.578.6400 -- You are currently subscribed to [email protected] as: [email protected] To unsubscribe, change settings or access archives, see http://www.ja-sig.org/wiki/display/JSG/cas-user
