| From: | "Zeugswetter Andreas SB SD" <ZeugswetterA(at)spardat(dot)at> |
|---|---|
| To: | "Yury Bokhoncovich" <byg(at)center-f1(dot)ru>, "Christopher Kings-Lynne" <chriskl(at)familyhealth(dot)com(dot)au> |
| Cc: | "Hackers" <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: [PATCHES] Big number of "unused" pages as reported by |
| Date: | 2002-09-06 07:33:03 |
| Message-ID: | 46C15C39FEB2C44BA555E356FBCD6FA4961E76@m0114.s-mxs.net |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
> I make a guess I've got this due to parallel running of a program making
> bulk INSERTs/UPDATEs into that table. Mmm...I need a way to avoid the big
> number of unused pages in such a case. LOCK TABLE?
Only UPDATEs and DELETEs (and rolled back INSERTs) cause unused pages.
The trick for other people was to run very frequent 'VACUUM goods;'
(like every 15 seconds) on tables when relatively few rows (in small tables)
where constantly beeing updated (e.g. counters/balances).
It might be sufficient in your case though to do the 'VACUUM goods;' after
every bulk UPDATE, like Christopher suggested. A concurrent vacuum won't
help if each bulk update is done in one single transaction.
Andreas
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Curt Sampson | 2002-09-06 07:53:12 | Re: Inheritance |
| Previous Message | Christopher Kings-Lynne | 2002-09-06 07:20:43 | Re: [PATCHES] Big number of "unused" pages as reported by VACUUM |