[Pgcluster-general] Maintaining Cluster DBs

[Josef.Bajada at go.com.mt]

Hi all,

Given I have just 2 nodes, each running load balancer, replicator and 
cluster db how could I achieve the following:

1. Determine which is the good and bad database, if I find that the 
databases are not synchronised. Is this ever supposed to happen?
2. How to bring down one of the nodes for maintenance (say upgrading of 
postgres and pgcluster)? Trying to stop postmaster normally does not seem 
to work for my test installation and I suspect that it is because the 
replicator from the other node is still connected. I end up killing it 
manually.
3. Lets say I want to upgrade postgres and pgcluster... will one node 
running say Postgres 8.1.5 with PGCluster 1.5rc13 be compatible with the 
upgraded node running Postgres 8.2 with PGCLuster 1.7rc3? If not, is it 
safe to upgrade PGCluster versions (for example between PGCluster 1.5rc10 
to 1.5rc13?)
4. Is there a way a load balancer or replicator can be queried the status 
of each Cluster DB, so that one can proactively know the health status of 
each node? (ideally through some interface that can be integrated 
programmatically into some monitoring process running on another system)

thanks,
Josef

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pgfoundry.org/pipermail/pgcluster-general/attachments/20070104/fe54b6f9/attachment.html