Sat Oct 20 18:21:16 PDT 2007
- Previous message: [Slony1-general] Postgres Upgrade Scenario Question
- Next message: [Slony1-general] Postgres Upgrade Scenario Question
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
David Rees wrote: > On 10/19/07, JP Fletcher <jpfletch at ca.afilias.info> wrote: >> David Rees wrote: >>> 1. Stop all slon daemons. >>> 2. Drop Pg 8.2 node from cluster. >>> 3. Stop Pg 8.2 >>> 4. Restart Pg 8.3 on the old port >>> 5. Update slon config for old port. >>> 6. Update sl_path info how??? >>> 7. Restart slon daemons >>> >>> Step 6 is where I get a bit unsure/lost. It looks like there is >>> connection information stored in the sl_path table and that should be >>> updated before starting up the slon daemons. >> What is it that you think should be updated? When you stop your 8.2 >> instance, and bring up the 8.3 instance on the port formerly occupied by >> 8.2, the conninfo in sl_path will match the 8.3 instance. > > Right, but the 8.2 instance is node 2, the 8.3 instance is node 3. > Step 2 includes dropping node 2 from the cluster. > > You could also ignore the fact that node 2 is going away for example, > and lets just say I wanted to change the port that node 3 is running > on... what's the "best practice" for doing so? When we do this sort of thing, we bring down both old and new postmasters, swap the ports, then restart. Then we reshape sl_path with slonik STORE PATH commands. And we wait to drop the old node until later. > > Updating sl_path is easy enough (only have 3 databases replicated on > this test setup, so that is 6 sl_path tables that need updating), I > just imagined that there would have been a more straight forward way > since everything else in slony is so straight forward. :-) > > -Dave
- Previous message: [Slony1-general] Postgres Upgrade Scenario Question
- Next message: [Slony1-general] Postgres Upgrade Scenario Question
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list