Thread: Some problem on the data base server !
Hello ! We had a database server on Mysql and we move to postgres last month. But there is problems that i don't understand ! When i runthe "top" command the load average is very high : 'load average: 3.23' or some times higher than that. And yestaerday i look to a 'top' and i found something strange : "CPU states: 95.1% user, 4.9% system, 0.0% nice, 59619.1% idle" >--------< is it normal ? can someone can give me some clues for decreasing the load ? i perform a VACUUM ANALYSE every day but it can't solve my problems. Thanx for answers. -- Mickaël JOUANNE Webmaster / Developpeur / Support Technique APR-Job & www.jobscout24.fr - Meet your future ___________ _______________________________________________ __________ HumanLine France SAS Tel: +33 1 42 44 40 73 32, rue Pierret Fax: +33 1 49 26 09 76 92200 Neuilly m.jouanne@humanline.fr -*- If Linux doesn't have the solution, you have the wrong problem -*-
On Mon, 20 Nov 2000, Mickaël Jouanne wrote: > We had a database server on Mysql and we move to postgres last month. Not enough. You need to specify the exact version, and also the version of your operating system. Let's assume Linux. > 'load average: 3.23' or some times higher than that. Who is responsible for that load ? Remember that on Linux, processes waiting for I/Os are also augmenting the load average. > "CPU states: 95.1% user, 4.9% system, 0.0% nice, 59619.1% idle" How is using the 95.1% of CPU ? > is it normal ? no, the 59619 value looks like a bug in top, or a mismatching kernel version.
On Mon, Nov 20, 2000 at 12:12:48PM +0100, Marc SCHAEFER wrote: > On Mon, 20 Nov 2000, Mickaël Jouanne wrote: > > > We had a database server on Mysql and we move to postgres last month. > > Not enough. You need to specify the exact version, and also the version of > your operating system. Let's assume Linux. > i have a Linux Mandrake release 7.1 (helium) Kernel 2.2.17 on a Bi-processor i686 and postgres 7.0.2 > > 'load average: 3.23' or some times higher than that. > > Who is responsible for that load ? Remember that on Linux, processes > waiting for I/Os are also augmenting the load average. > there is only postmaster on the server so it's just query that make this load ! > > "CPU states: 95.1% user, 4.9% system, 0.0% nice, 59619.1% idle" > > How is using the 95.1% of CPU ? > i don't know ! > > is it normal ? > > no, the 59619 value looks like a bug in top, or a mismatching kernel > version. > -- Mickaël JOUANNE Webmaster / Developpeur / Support Technique APR-Job & www.jobscout24.fr - Meet your future ___________ _______________________________________________ __________ HumanLine France SAS Tel: +33 1 42 44 40 73 32, rue Pierret Fax: +33 1 49 26 09 76 92200 Neuilly m.jouanne@humanline.fr -*- If Linux doesn't have the solution, you have the wrong problem -*-
On Mon, 20 Nov 2000, Mickaël Jouanne wrote: > there is only postmaster on the server so it's just query that make this > load ! > > How is using the 95.1% of CPU ? who, I meant. ie is PostgreSQL the top CPU eater ? What gives the `free' command ?
On Mon, Nov 20, 2000 at 02:55:25PM +0100, Marc SCHAEFER wrote: > On Mon, 20 Nov 2000, Mickaël Jouanne wrote: > > > there is only postmaster on the server so it's just query that make this > > load ! > > > > How is using the 95.1% of CPU ? > > who, I meant. ie is PostgreSQL the top CPU eater ? > the postmaster are the more CPU eater : 3:00pm up 15 days, 23:32, 5 users, load average: 0.81, 1.15, 1.10 287 processes: 284 sleeping, 3 running, 0 zombie, 0 stopped CPU states: 20.5% user, 9.1% system, 0.0% nice, 70.2% idle 15928 postgres 10 0 14156 13M 13612 R 0 5.9 0.7 0:00 postmaster 15929 postgres 14 0 6712 6712 6212 R 0 3.0 0.3 0:00 postmaster 3591 postgres 0 0 19660 19M 18576 S 0 0.5 0.9 0:07 postmaster > What gives the `free' command ? free total used free shared buffers cached Mem: 2009560 1758128 251432 1813748 895052 494700 -/+ buffers/cache: 368376 1641184 Swap: 1019976 0 1019976 -- Mickaël JOUANNE Webmaster / Developpeur / Support Technique APR-Job & www.jobscout24.fr - Meet your future ___________ _______________________________________________ __________ HumanLine France SAS Tel: +33 1 42 44 40 73 32, rue Pierret Fax: +33 1 49 26 09 76 92200 Neuilly m.jouanne@humanline.fr -*- If Linux doesn't have the solution, you have the wrong problem -*-