my mail server is drowning

[Jeff A. Earickson]

On Wed, 7 May 2003, Julian Field wrote:

>
> Correct. But make sure you *don't* have
> Always Include SpamAssassin Report = yes
> set, otherwise it will spamassassinate everything just to produce the
> report you asked for.
>
Yes, turned off.

> >BTW, I turned off auto-whitelist (per Julian's suggestion) and things
> >are holding together a little better today.  MailScanner still has a
> >big memory footprint (36 M).  I reduced the number of children from 6
> >to 4, still a lot of memory page activity.
>
> 4 * 36MB = 144MB.
> So where is the rest of your 2048 MB going? Not on MailScanner, according
> to your figure.

Here is what the output of "top -u -osize" shows (first few lines):

load averages:  5.62,  5.20,  5.57                                     11:42:20
95 processes:  85 sleeping, 8 running, 2 on cpu
CPU states:  0.0% idle, 64.2% user, 35.8% kernel,  0.0% iowait,  0.0% swap
Memory: 2048M real, 1296M free, 644M swap in use, 1987M swap free

   PID    UID   THR PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND
   199     76     7  58    0  121M  118M sleep    6:33  3.03% named
 26225      0     1  52    0   38M   17M run      0:01  2.59% MailScanner
 17182      0     1  33    0   38M   37M sleep    2:25  0.99% MailScanner
 26259      0     1  52    0   38M 9072K run      0:00  0.55% MailScanner
 17199      0     1  51    0   37M   36M run      2:32  1.67% MailScanner
 26269      0     1  52    0   37M 7176K cpu/0    0:00  0.19% MailScanner
 17230      0     1  25    0   37M   36M sleep    2:32  1.05% MailScanner
 17259      0     1  43    0   37M   36M sleep    2:34  0.78% MailScanner
 26250      0     1  51    0   37M   24M run      0:00  1.42% MailScanner
   586      0    52  30    0   25M   23M sleep    0:42  0.24% dced
  8944      0     8  60    0   12M 7208K sleep    0:00  0.00% sshd
  8987   9897     3  58    0   12M 4136K sleep    0:00  0.00% sshd
   758      0     9  34    0   12M 6544K sleep    0:00  0.00% sshd
   588      0     9  60    0   12M 6312K sleep    0:00  0.00% sshd
   386      0     8  18    0   12M 6168K sleep    0:00  0.00% sshd
   687      0     8  34    0   12M 6312K sleep    0:00  0.00% sshd
   610  13462     3  58    0   12M 3912K sleep    0:00  0.00% sshd
   808  13462     3  58    0   12M 4008K sleep    0:00  0.00% sshd
   704  13462     3  58    0   12M 3936K sleep    0:00  0.00% sshd
   415  13462     3  58    0   12M 3832K sleep    0:02  0.00% sshd
 17181      0     1  57    0   11M 2720K sleep    0:00  0.00% MailScanner
   780      0    14  58    0   10M 8112K sleep    0:01  0.00% cdsclerk
 26264      0     8  28    0   10M 7056K sleep    0:00  0.48% popper
 26235  17491     8  58    0 9976K 6984K sleep    0:00  0.70% popper
 26220  17624     8  41    0 9976K 6984K sleep    0:00  0.43% popper
 26236  18638     8  58    0 9968K 6976K sleep    0:00  0.60% popper

I don't think it is lack of memory, just a high number of processes and
a lot of process switching that is causing the "pi" number to be high
in "vmstat".

--- Jeff