Hello all,

I am running qmail1.03 on SunOS5.7 single processor, single disk
machine. This is an incoming qmail box that processes around 75,000
messages a day. It does no local delivery, it simply relays all mail to a
certain host with a line like the following in smtproutes:

:host.domain

What has started to happen today is that receiving mail is slow, between
the '.' and 'ok' from qmail (got that from telnet to port 25). Even slower
is its sending, which is about a message every 5 or 10 seconds. The other
noticable problem is that there are thousands of qmail-queue processes
running concurrently in the process table. Most of the processes are
fairly recent (started within the last couple of minutes), but some are
several hours old. Here is a sample of ps -ef:

  qmailq 20471     1  0 15:51:29 ?        0:00 bin/qmail-queue
  qmailq 19623     1  0 15:51:10 ?        0:00 bin/qmail-queue
  qmailq 20899     1  0 15:51:40 ?        0:00 bin/qmail-queue
  qmailq 21515     1  0 15:51:54 ?        0:00 bin/qmail-queue
  qmailq 17269     1  0 15:50:17 ?        0:00 bin/qmail-queue
  qmailq 19408 19385  0 15:51:06 ?        0:00 bin/qmail-queue

There a thousands of these processes, and as you can see, some of them
have init as a parent, and some have an actual qmail-smtpd.

The queue is growing on this incoming mail server because it is sending
mail so infrequently/slowly.

Can somebody please give me some advice on where to start.

Oh, iostat looks like this:

                              extended device statistics
  r/s  w/s   kr/s   kw/s wait actv wsvc_t asvc_t  %w  %b device
  0.0  0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c0t0d0s0
  0.0  0.2    0.0    1.3  0.0  0.1    0.0  311.2   0   5 c0t0d0s1
  0.0  0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c0t0d0s2
  0.0  0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c0t0d0s3
  1.3 209.8   12.5 1281.9  0.0  6.3    0.0   29.6   0  99 c0t0d0s4
  0.0  0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0 c0t0d0s6
  0.0  0.1    0.0    0.4  0.0  0.1    0.0  628.6   0   5 c0t0d0s7
  0.0  0.0    0.0    0.0  0.0  0.0    0.0    0.0   0   0
cobalt:vold(pid207)

s4 is mounted as /var/qmail. Obviously pretty busy 99%! I think this 99%
busy is a result of these qmail-queue porcess, however I am hesitant to
believe that this problem is as simple as the disk is maxing out and thus
the hanging qmail-queue porcesses. Especially because this box has been
able to handle the load thus far, and I do not detect any new extremely
large influx of mail.

I would be tremendously grateful for any advice tackling this problem.

Brandon

Reply via email to