Thu Apr 23 07:19:13 PDT 2009
- Previous message: [Slony1-general] will truncating a table on the master be properly handled by slony on the slave?
- Next message: [Slony1-general] will truncating a table on the master be properly handled by slony on the slave?
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Thu, Apr 23, 2009 at 10:13:41AM -0400, Geoffrey wrote: > So, the solution to this is, I assume 'execute script?' Well, maybe. The point of truncating is usually to do something fast, cheap, and easy. EXECUTE SCRIPT entails a heavyweight lock. Often, there _is_ no real way to do this easily. Since the table is guaranteed to be empty after a truncate, of course, you can just take it out of replication, truncate it, and re-add it: the latter should be very fast on an empty table. It's still a locking operation, though. If you have a table you want to truncate a lot, the better bet is to use inheritance and a couple different underlying tables, so that you have a table that is guaranteed "out of service" on all nodes. Then you can truncate that table everywhere without fear. This is a loaded foot-gun, however, so be careful. > Also, what is the best way to get these tables back in sync without > trashing slony and starting over? > > Can we drop the tables from replication and re-add them? Yes. A -- Andrew Sullivan ajs at crankycanuck.ca
- Previous message: [Slony1-general] will truncating a table on the master be properly handled by slony on the slave?
- Next message: [Slony1-general] will truncating a table on the master be properly handled by slony on the slave?
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list