[PERFORM] select count(*) performance (vacuum did not help)

2007-09-24 Thread Gábor Farkas
hi, i have the following table: CREATE TABLE "main_activity" ( "id" serial NOT NULL PRIMARY KEY, "user_id" integer NOT NULL, "sessionid" varchar(128) NOT NULL, "login" timestamp with time zone NOT NULL, "activity" timestamp with time zone NOT NULL, "logout" timestamp with

Re: [PERFORM] select count(*) performance (vacuum did not help)

2007-09-24 Thread Gábor Farkas
Heikki Linnakangas wrote: Gábor Farkas wrote: - but in the past, the cron-job was not in place, so the table's size grew to around 80 entries (in around 80 days) That bloated your table, so that there's still a lot of empty pages in it. VACUUM FULL should bring it back to a

Re: [PERFORM] select count(*) performance (vacuum did not help)

2007-09-24 Thread Gábor Farkas
Heikki Linnakangas wrote: Gábor Farkas wrote: if i do a full vacuum to that table only, will the database still serve data from the other tables at a normal speed? Yes. The extra I/O load vacuum full generates while it's running might disrupt other activity, though. i see. will i ac

[PERFORM] postgresql on NFS.. recommended? not recommended?

2007-10-08 Thread Gábor Farkas
hi, is it recommended to run a postgresql server on a nfs-share (gigabit-network)? so basically i have a NAS + a database-server, and wonder if i should put the database on local hard-drives in the db-server, or on the NAS, and mount it using NFS on the database-server. i haven't investigated

[PERFORM] autovacuum: recommended?

2007-11-16 Thread Gábor Farkas
hi, we are moving one database from postgresql-7.4 to postgresql-8.2.4. we have some cronjobs set up that vacuum the database (some tables more often, some tables less often). now, in pg82, there is the possibility of using the autovacuum. my question is: is it recommended to use it? or in o

[PERFORM] postgresql uses Hash-join, i need Nested-loop

2009-08-23 Thread Gábor Farkas
hi, i have a query that uses a Hash-Join, but it would be faster with Nested-Loop, and i don't know how to persuade postgresql to do it. details: postgresql-8.2 + tsearch2 i have 2 tables, one for people, and one that does a many-to-many link between people: CREATE TABLE personlink ( id in