[imp] high load average with postgresql

Matthew M Carroll mmc2@csd.uwm.edu
Tue, 24 Apr 2001 14:31:49 -0500 (CDT)


Im sorry I was a bit vague 
My machine is a dual 750 with 1Gig of ram. Im running suse 7.1 with 2.4.0
I am using a mylex accelraid 170 with a 4 disk raid 5 set. at most I will
have 150 or so users on the box averaging 3000 unique hits a day and 6000
total hits a day. I am running horde/imp 2.2.4/1.2.4 I am running postgres
7.0.3, and the imap server is UW imap on 3 other boxes which imp connects
to.

My postmaster startup parameters are 

/usr/bin/postmaster -p 5432 -D /var/lib/pgsql/data -A 1 -B 768 -b
/usr/bin/postgres -i -N 384 

I guess I am wondering if the -F parameter on postgres will help that
much?

Here is a sample top output

  2:26pm  up 21:38,  3 users,  load average: 20.83, 14.16, 17.22
208 processes: 191 sleeping, 17 running, 0 zombie, 0 stopped
CPU0 states: 41.5% user, 56.1% system,  0.0% nice,  1.3% idle
CPU1 states: 37.4% user, 60.4% system,  0.0% nice,  1.1% idle
Mem:   900028K av,  667444K used,  232584K free,       0K shrd,   13792K
buff
Swap:  273088K av,       0K used,  273088K free                  315368K
cached

  PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
 6135 postgres  17   0  8544 8544  8088 S     7.4  0.9   0:03 postmaster
 5953 postgres  17   0  8544 8544  8088 R     7.2  0.9   0:06 postmaster
 5999 postgres  16   0  8548 8548  8088 R     7.0  0.9   0:06 postmaster
 6159 postgres  14   0  8544 8544  8088 R     7.0  0.9   0:02 postmaster
 5998 postgres  14   0  8544 8544  8088 R     6.8  0.9   0:05 postmaster
 6003 postgres  11   0  8544 8544  8088 S     6.8  0.9   0:05 postmaster
 6149 postgres  14   0  8548 8548  8088 R     6.6  0.9   0:02 postmaster
 5966 postgres  15   0  8548 8548  8088 S     6.3  0.9   0:06 postmaster
 6136 postgres  14   0  8544 8544  8088 R     6.3  0.9   0:03 postmaster
 5942 postgres  12   0  8548 8548  8084 S     6.1  0.9   0:06 postmaster
 5951 postgres  13   0  8544 8544  8088 S     6.1  0.9   0:06 postmaster
 5967 postgres  14   0  8544 8544  8084 S     6.1  0.9   0:05 postmaster

As far as the database goes its a newly converted mysql database that has
been vacuumed yesterday. 

Again any info or ideas would be appreciated, sorry for being so vague I
don't know what I was thinking.

Thanks, 

 On Tue, 24 Apr 2001, Rich Lafferty wrote:

> On Tue, Apr 24, 2001 at 11:07:41AM -0500, Matthew M Carroll (mmc2@csd.uwm.edu) wrote:
> > I recently changed my horde/imp installation over to postgresql, and now
> > under the same conditions I have a load average of 10+ where as before I
> > would have a load average of 2 and under. this is a smp machine. If anyone
> > has any ideas on how to tune this better I would appreciate it. 
> 
> Well, I don't know what "before" was. What have you tried that hasn't
> worked? What's the machine? What else is on the machine? What version
> of Postgres? What version of Horde and IMP? What operating system?
> What processes are actually taking up the load? How many postgres
> processes are running? What parameters are you starting Postgres with?
> How are you connecting to Postgres? What sort of user load is on the
> machine? How often do you VACUUM and VACUUM ANALYZE?
> 
> It's impossible to know what the problem is without any information at
> all.
> 
>   -Rich
> 
> -- 
> ------------------------------ Rich Lafferty ---------------------------
>  Sysadmin/Programmer, Instructional and Information Technology Services
>    Concordia University, Montreal, QC                 (514) 848-7625
> ------------------------- rich@alcor.concordia.ca ----------------------
> 
> -- 
> IMP mailing list: http://horde.org/imp/
> Frequently Asked Questions: http://horde.org/faq/
> To unsubscribe, mail: imp-unsubscribe@lists.horde.org
> 
> 
> 


Matthew M Carroll
mmc2@csd.uwm.edu