> -----Original Message-----
> From: [EMAIL PROTECTED] 
> [mailto:[EMAIL PROTECTED] On Behalf Of 
> Krzeminski, Damian (BL60:9D30)
> Sent: Wednesday, October 22, 2008 9:03 AM
> To: [EMAIL PROTECTED]
> Subject: [sipX-dev] alarm server in a cluster
> 
> I was reviewing some new issues and future request related to 
> Alarm Server management and I have a question:
> 
> Does it really make sense to run a separate instance of an 
> alarm service on each server in the cluster?
> 
> Why can't all services report the alarms to a single instance 
> of the server. They can be easily configured that way, and it 
> might make alarm reporting and aggregating simpler.
> Thoughts?
> D.

It was originally designed that way, and some of the existing config
reflects that (for example, ALARM_SERVER_URL in domain-config).
However, way back in June, Scott suggested that it would be best left
distributed (http://list.sipfoundry.org/archive/sipx-dev/msg11878.html)
and we agreed yesterday in a conversation that that is still desired.

I am just starting to clean up the ALARM_SERVER_URL, which does NOT
belong in domain-config if we have distributed alarm servers, so let's
agree on what we want.

Single instance of alarm server on master (or on config_host)
+ easier to display alarms - they're all in one place
o ALARM_SERVER_URL stays in domain-config, with the master hostname.
(config_host may eventually be duplicated)
- all clients need to pull dest out of domain-config (currently scripts
and Java use localhost)
- we lose alarms if the master goes down, or the network

Alarm server in each node (part of each supervisor)
+ alarms are logged even if master or network goes down
- display of alarms must be aggregated

Carolyn
_______________________________________________
sipx-dev mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-dev
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-dev

Reply via email to