On Mon, March 24, 2008 4:13 pm, Andrew Sullivan wrote: > On Mon, Mar 24, 2008 at 11:02:18AM +0200, Henry wrote: >> Hello all, >> >> During an initial replication from a master to many slaves, why does >> slony >> perform an ANALYZE after the COPY? For paltry tables this is OK, but >> for > > For the same reason you need to ANALYSE after a restore from backup.
Hmm, ok. I was under the mistaken impression that ANALYZE after COPY was like fsck after mkfs ... ie, unnecessary. I re-read the PG docs, and yup, you're right. ANALYZE after bulk loading is always a good idea for performance. >> massive (multiple GB) tables, this is something which takes frigging >> *days* per table. > > ?! If so, you have something wrong. ANALYSE shouldn't take that long. > It samples the table. OK, maybe not days per table, but what feels like hundreds of hours per bloody table... ;-) > >> Even if there is a valid reason to analyze after a copy, is there a way >> to >> disable this behavior (short of hacking the src)? The docs don't tell >> me >> much. > > Nope. Your performance on replication will be totally miserable without > it. > You MUST do it. Right you are. Thanks for the comments. _______________________________________________ Slony1-general mailing list [email protected] http://lists.slony.info/mailman/listinfo/slony1-general
