Fri Oct 15 14:52:25 PDT 2004
- Previous message: [Slony1-general] recovery
- Next message: [Slony1-general] recovery
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On 10/14/2004 5:06 PM, David Parker wrote: > Given 2 nodes A and B in a slony replication cluster, with A as the > provider for a single set, if at some point I need to failover to B, > what are my options for eventually returning A to provider status? Once > we've failed over to B, we eventually want to get A back to being the > main guy. > > It looks like the first step is to set up a fresh schema in A, and > subscribe it to B. I thought at first I would want to populate it first > with a pg_dump from B, but I see that the ENABLE_SUBSCRIPTION logic > truncates the target table before copying to it. You might want to read this: http://gborg.postgresql.org/project/slony1/genpage.php?howto_failover > > Will simply subscribing the new A to B result in all existing B data > geting copied over? If so, how does this compare performance-wise with > using pg_dump? (I realize this comparison is apples vs. oranges, and > really a comparision between table copy and pg_dump, hence not > slony-specific, but if anybody has any experience with this I'd > appreciate hearing about it). Since you are setting up just a new replica, it will probably take as long as it originally took to build B. That assumes that the two machines are similar. The current bottleneck of the Slony copy process is memory if there are a lot of indexes on the tables. > > Assuming I'm able to fully load A with data from B, is it possible to > switch them such that A becomes the provider and B goes back to receive > mode? This is complicated if updates are continuing to B, of course, but > I wonder if it is possible at all. That's the whole point of the switchover capabilities of Slony. > > Finally, what query would I execute against 2 nodes to know if their > data is in sync (at that moment)? In 1.0.3 comes a new view sl_status, that tells you on the origin how far all subscribers have confirmed events or how far behind they are. Jan > > Thanks in advance for any answers/corrections to any of my numerous > questions! > > - DAP > ====================================================== > David Parker Tazz Networks (401) 709-5130 > > _______________________________________________ > Slony1-general mailing list > Slony1-general at gborg.postgresql.org > http://gborg.postgresql.org/mailman/listinfo/slony1-general -- #======================================================================# # It's easier to get forgiveness for being wrong than for being right. # # Let's break this rule - forgive me. # #================================================== JanWieck at Yahoo.com #
- Previous message: [Slony1-general] recovery
- Next message: [Slony1-general] recovery
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list