Hi,
We're looking at a clustered solution and need to choose between standby and HA cluster.
Regarding a standby cluster:
In a standby cluster, where the database is in standby mode (not accessible), does the standby Alfresco server run, waiting for the DB to come available, or can Alfresco only be started once the database is made active?
Can we confirm that index tracking syncs the Lucene indexes against the database, meaning that a standby server cannot use index tracking (since the DB isn't active).
When Alfresco is started and the indexes are stale, does the server become immediately available, or does the index restoration need to complete first? We need 24x7 read/write access, so switching to read-only mode for five minutes each day to take index backups is not likely to be acceptable. Since we'd fail over in rare circumstances, we can expect the indexes to be extremely stale.
Regarding a HA cluster:
Is there any built-in monitoring for the replicating content store's asynchronous mode? For example, if the WAN between primary and secondary were down for 30 minutes, is there a way to get alerts telling us that a failover would risk significant data loss?
If we used the synchronous replicating content store, and the server cannot save the data to the remote server's repository (perhaps a WAN outage), would this cause the transaction to abort? i.e. would loss of the WAN connection between the clustered nodes cause full loss of the cluster? I expect it would so I'm leaning towards async replicating content stores, hence the monitoring question.
Many thanks,
Brett
