Please review the man page for vmstats. You may find it helpful to buy
and read any of the books on UNIX performance tuning (the ones on
Solaris tuning would be most usefull). Your swap file is in serious
contention with your processes since you are badly over committed. You
might also want to reduce the number of processes doing qmail injection.
You will find your performance improves dramatically.

Jim Arnott wrote:
> 
> Hi,
> 
> I meant the "cat >> out.file < 1000byte.in" was done in a 1000 time loop.
> 
> #vmstat 1 (during inject test):
> 
> Virtual Memory Statistics: (pagesize = 8192)
>   procs    memory         pages                          intr        cpu
>   r  w  u  act  free wire fault cow zero react pin pout  in  sy  cs  us  sy  id
>   2120 29  8705  18K 4026 1141  453  343    0  382    0 118  3K  1K   2  24  74
>   2120 31  8790  18K 3977  670  260  216    0  224    0 125  2K  1K   2  15  83
>   4119 29  8822  18K 3935  989  395  291    0  321    0  96  3K  1K   4  27  69
>   2120 30  8825  18K 3936 1030  418  309    0  375    0 118  3K  1K   2  27  71
>   2120 31  8856  18K 3953 1049  426  309    0  360    0 109  4K  1K   4  29  67
>   2120 30  8833  18K 3928 1231  501  354    0  436    0 121  2K  1K   3  22  75
>   4119 29  8836  18K 3932  667  258  216    0  221    0 120  2K  1K   2  18  80
> 
> 256 Meg mem, Latest 10 Gig SCSI
> 
> -jim
> 
> >
> > There is something very wrong here. Cat'ing two blocks of data should
> > take milliseconds. Are you out of memory (does vmstat show paging?)?
> > When your cat is << .1 seconds or so then rerun your tests. Wait a
> > minute, did you run the cat WHILE you were trying to do the deliveries?
> > If so, then that is about right as qmail is beating on your hard disk.
> >
> > Your qmail-inject seem like they are off by about a factor of 10 or less
> > (since qmail-send isn't running), but that is just a swag (assuming
> > about a hundred other things like a properly configured file system, no
> > swapping/page stealing (sufficent memory), and a current-technology SCSI
> > non-IDE drive).
> >
> > Jim Arnott wrote:
> > >
> > > Hi,
> > >
> > > I'm experimenting with qmail's throughput. It approximates what
> > > we want to do with it. (send mail from a mail database through qmail-inject)
> > > I want to see how fast qmail will queue messages.
> > >
> > > I have a  ~433 Mhz Alpha.
> > >
> > > Experiment:
> > >
> > >         script that runs qmail-inject 1000 times
> > >         with a 1000 byte body
> > >         Qmail-send is not running
> > >
> > > Result:
> > >         takes 73 seconds (13.7/sec)
> > >
> > > This seems a little slow to me. The system cpu is 70% idle.
> > >
> > > Same experiment with "cat >> out.file < 1000byte.in" takes 4 seconds.
> > >
> > > Is this normal ? Any ideas on how to speed it up.
> > >
> > > thanks,
> > > -jim
> >
> > --
> > Daemeon Reiydelle
> > Systems Engineer, Anthropomorphics Inc.
> > [EMAIL PROTECTED]

-- 
Daemeon Reiydelle
Systems Engineer, Anthropomorphics Inc.
[EMAIL PROTECTED]

Reply via email to