Our dual opteron has been performing well for many
weeks now (after some simple tuning) when all of a sudden the queries have
slowed right down!
ie: DUAL 246 OPTERON:
select count(*) from job_archieve; - Time: 107.24 ms
explain analyse select count(*) from
job_archieve;
Aggregate (cost=2820.50..2820.50 rows=1 width=0) (actual time=153.53..153.53 rows=1 loops=1) -> Seq Scan on job_archieve (cost=0.00..2789.20 rows=12520 width=0) (actual time=1.39..132.98 rows=12520 loops=1) Total runtime: 153.74 msec Time: 156.94 ms CRAPPY AMD ATHLON XP 1700+:
select count(*) from job_archieve; - Time: 23.30 ms
explain analyse select count(*) from
job_archieve;
Aggregate (cost=2816.50..2816.50 rows=1
width=0) (actual time=133.83..133.84 rows=1 loops=1)
-> Seq Scan on job_archieve (cost=0.00..2785.20 rows=12520 width=0) (actual time=0.02..72.64 rows=12520 loops=1) Total runtime: 133.92 msec Time: 134.79 ms The ratio of these simple query times is about
accurate for most queries performed on the same database on the different
machines... Any ideas what may have suddenly caused this and where to start
troubleshooting??? Both dbs have already been fully vacuumed.
The opteron is going to get a overhaul (4 port raid
going in, fresh install of freebsd, postgres etc) but would be handy to know for
future reference in case this happens again....
(ps, yes i know archive is not spelt archieve
;)
Cheers!
Dave. |
- [PERFORM] gah! sudden slowdown?? SpaceBallOne
- Re: [PERFORM] gah! sudden slowdown?? Steinar H. Gunderson