On 02/05/2013 04:48 PM, Christopher Browne wrote:
> It's probably also conceivable to use PITR to this end, checking the
> end XID on each database node, and setting things up to recover to
> that point.  But I rather expect that filesystem/disk array snapshots
> will be the easiest way to do this.

Is there a function or query we could use to verify that the nodes are
all in sync as far as the slony replicated data?

If so, I'm thinking we could then do something like:

On backup:
1) stop all activity on the cluster
2) check the nodes are all in sync
   -> if not wait, check again
3) once all in sync
   -> back up data-only from slon schema on each node
   -> back up user tables data-only on master node
4) start up slony, recommence activity

On restore:
1) stop all activity on the cluster
2) stop all the slons
3) disable the slony triggers
4) truncate all slony and user tables
5) reload all nodes from previous backups
   -> slony data from local node backup
   -> user data from master node backup
6) re-enable triggers
7) start all the slons, recommence activity

Does this sound feasible?

Thanks,

Joe


-- 
Joe Conway
credativ LLC: http://www.credativ.us
Linux, PostgreSQL, and general Open Source
Training, Service, Consulting, & 24x7 Support