Scott Lawrence wrote: >> Scott Lawrence wrote: > >>> I'm interested in opinions on which services that are not currently >>> redundant that users thing should be. If you could pick just one >>> service for us to add HA to, what would it be? > > On Tue, 2009-07-14 at 14:41 +0100, Keith Gearty wrote: > >> SipXconfig. The ability to be able to administer the system through >> the web interface while the primary server is down would make me >> seriously consider using HA. > > That one has some very interesting implications... now that we've got > the configuration data centralized into sipXconfig properly (a major > element of 4.0), it's not as hard as it used to be. > > Theoretically, we could replicate configuration from the master (the > original system) to some standby configuration server, which could then > be used if/when the original failed. The interesting questions all have > to do with detecting when it's appropriate to allow the backup server to > make a change, and which server is believed to 'own' the configuration. > > What would you expect to happen when the original master came back up? > > How automatic would you expect the change to be? > > Some of the tricky questions revolve around whether the active master > has failed or is actually up but unreachable (some elements of the > cluster are partitioned from others by some network failure). > >
One way I can imagine that is that sipXconfig runs separately from primary (and other servers). And the redundancy is provided by the servlet/application container and possibly by the DB server. The question is how to add all that flexibility without making it all too complicated to install and maintain. D. _______________________________________________ sipx-users mailing list sipx-users@list.sipfoundry.org List Archive: http://list.sipfoundry.org/archive/sipx-users Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users sipXecs IP PBX -- http://www.sipfoundry.org/
