Re: [PERFORM] High load average with PostgreSQL 7.4.2 on debian/ibm eserver.
[EMAIL PROTECTED] wrote: Whole config is available here: http://ludojad.itpp.pl/~eleven/pg-high-load.conf effective_cache_size = 4000 # typically 8KB each #random_page_cost = 4 # units are one sequential page fetch cost #cpu_tuple_cost = 0.01 # (same) #cpu_index_tuple_cost = 0.001 # (same) #cpu_operator_cost = 0.0025 # (same) These values are too higher for your hardware, try to execute the explain analyze for the queries that are running on your box and repeat it lowering these values, I bet postgres is running seq scan instead of an index scan. These are the value that I use for a configuration closer to your: effective_cache_size = 2 random_page_cost = 2.0 cpu_tuple_cost = 0.005 cpu_index_tuple_cost = 0.0005 cpu_operator_cost = 0.0025 last question, do you use the autovacuum daemon ? If no = you have to use it If yes = did you apply the patch that will not fail with big tables like yours ? if you can post the autovacuum daemon log ( last lines ). Regards Gaetano Mendola ---(end of broadcast)--- TIP 8: explain analyze is your friend
Re: [PERFORM] High load average with PostgreSQL 7.4.2 on debian/ibm eserver.
On Tue, Jun 29, 2004 at 09:17:36AM -0700, Marc wrote: Performance issue, I'm experiencing here, is somewhat weird - server gets high average load (from 5 up to 15, 8 on average). Standard performance monitoring utilities (like top) show that CPUs are not loaded (below 20%, often near zero). So ... you never actually say what the performance issue you experience is. Having a high load average is not necessarily a performance issue. Well, if the server's CPUs are idle and the machine is starting to hog itself, one can suspect something bad going on. What is it that you want to fix? Basically, I'm wondering if I'm already on the edge of performance capabilities of this machine/configuration, or maybe there's some abnormal behaviour happening (which could be noticed by somebody from this mailing list, hopefully). In particular - could someone tell me if those iostat values can tell if I'm close to upper performance boundary of fast SCSI (Ultra 320, 15k RPM) disks? -- 11. ---(end of broadcast)--- TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED]
Re: [PERFORM] High load average with PostgreSQL 7.4.2 on debian/ibm eserver.
Eleven, In particular - could someone tell me if those iostat values can tell if I'm close to upper performance boundary of fast SCSI (Ultra 320, 15k RPM) disks? It's quite possible that you need to improve your disk array; certainly I would have spec'd a lot more disk than you're using (like raid 0+1 with 6 disks or RAID 5 with seven disks). However, there's the other end as well; it's quite possible that your queries are doing seq scans and other disk-intensive operations that could be avoided. Have you analyed this at all? -- -Josh Berkus Aglio Database Solutions San Francisco ---(end of broadcast)--- TIP 2: you can get off all lists at once with the unregister command (send unregister YourEmailAddressHere to [EMAIL PROTECTED])