I am by no means a expert on this, but I believe that by default with
only one node left, the cluster goes into Read Only and won't accept
writes, so in theory, with a two node cluster you would just have to
bring back the failed node. You may however, just to be on the safe
side, bring the failed node up with -R since, well, who knows if you had
file corruption when the node went south.

The relevant cluster.conf setting is 
<When_Stand_Alone> read_only </When_Stand_Alone>

If you've changed that, then yes, you absolutely have to bring back a
failed node with -R as the replicator does not keep track of
transactions that have occurred.

On Thu, 2008-08-28 at 09:04 -0500, David Schlenk wrote: 
> I'm new to pgcluster and have a question about how its supposed to work.
> 
> In my test cluster, I have two DB/cluster nodes. I have two cascaded
> replicators working using RLOG (which, btw, it would be nice to add a blurb
> to the install document about how to do this... I had to dig through the
> mailing list to figure out how to configure it properly) and a couple load
> balancers (although the intent is to use only one at a time and have the
> second kill the first if it detects a problem and then take its IP). I've
> been testing recovery, and my question is this:
> If I shut down one of the DBs, then perform a write, then bring that DB back
> up, should the replicator notice that it's behind and push any pending
> writes to it? Or do you have to start with the -R option for recovery to
> happen? My concern is that if one DB goes down, some writes happen, then the
> 2nd goes down, if you bring up the node that went down first, then bring up
> the other node in recovery mode you'll lose any writes that happened after
> the first DB went down.

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
Pgcluster-general mailing list
[email protected]
http://pgfoundry.org/mailman/listinfo/pgcluster-general

Reply via email to