E-Mail Server Load

fukawi2 · 02-28-2007, 12:00 AM

Hi all,

I'm currently running a mail server (qmail) which seems to have a very high load average. I want to try and find where the bottleneck is (I suspect disk I/O, although I'm running RAID)

The server is a Dual Pentium III 1.2Ghtz w/ 1gb RAM and 3 x 36gb 10k SCSI drives in RAID 5. The disks are formatted as ext3, I'm wondering if XFS / JFS / ReiserFS may be more suitable for this task...?

There is still 32mb RAM free, and the CPU's are generally idling. The only other thing I can think of is disk I/O. Can anyone help me with how to find what's causing the bottleneck?

Code:

top - 16:55:23 up 22 days, 58 min,  1 user,  load average: 0.94, 0.95, 1.19
Tasks: 115 total,   1 running, 114 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.3%us,  1.0%sy,  0.0%ni, 95.7%id,  2.8%wa,  0.2%hi,  0.0%si,  0.0%st
Mem:   1034588k total,  1001804k used,    32784k free,    33872k buffers
Swap:  2096440k total,       92k used,  2096348k free,   809800k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 1616 phillips  15   0  2156 1024  792 R    2  0.1   0:08.70 top
    1 root      15   0  2032  640  556 S    0  0.1   0:01.63 init [3]
    2 root      RT   0     0    0    0 S    0  0.0   0:22.09 [migration/0]
    3 root      34  19     0    0    0 S    0  0.0   0:00.89 [ksoftirqd/0]
    4 root      RT   0     0    0    0 S    0  0.0   0:00.03 [watchdog/0]
    5 root      RT   0     0    0    0 S    0  0.0   0:14.73 [migration/1]
    6 root      34  19     0    0    0 S    0  0.0   0:00.47 [ksoftirqd/1]
    7 root      RT   0     0    0    0 S    0  0.0   0:00.00 [watchdog/1]
    8 root      10  -5     0    0    0 S    0  0.0   0:02.26 [events/0]
    9 root      10  -5     0    0    0 S    0  0.0   0:00.56 [events/1]
   10 root      17  -5     0    0    0 S    0  0.0   0:00.03 [khelper]
   11 root      17  -5     0    0    0 S    0  0.0   0:00.06 [kthread]
   53 root      12  -5     0    0    0 S    0  0.0   0:17.38 [kblockd/0]
   54 root      10  -5     0    0    0 S    0  0.0   0:03.51 [kblockd/1]
   55 root      15  -5     0    0    0 S    0  0.0   0:00.00 [kacpid]
  122 root      15  -5     0    0    0 S    0  0.0   0:00.00 [cqueue/0]
  123 root      15  -5     0    0    0 S    0  0.0   0:00.00 [cqueue/1]
  124 root      15  -5     0    0    0 S    0  0.0   0:00.00 [ksuspend_usbd]
  127 root      15  -5     0    0    0 S    0  0.0   0:00.00 [khubd]
  129 root      10  -5     0    0    0 S    0  0.0   0:00.00 [kseriod]
  151 root      10  -5     0    0    0 S    0  0.0  12:34.21 [kswapd0]
  152 root      15  -5     0    0    0 S    0  0.0   0:00.00 [aio/0]
  153 root      15  -5     0    0    0 S    0  0.0   0:00.00 [aio/1]
  306 root      11  -5     0    0    0 S    0  0.0   0:00.00 [kpsmoused]
  340 root      15  -5     0    0    0 S    0  0.0   0:00.00 [scsi_eh_0]
  341 root      20  -5     0    0    0 S    0  0.0   0:01.33 [aacraid]
  342 root      10  -5     0    0    0 S    0  0.0  16:27.29 [kjournald]
  370 root      11  -5     0    0    0 S    0  0.0   0:00.00 [kauditd]
  404 root      18  -4  2116  656  452 S    0  0.1   0:02.46 /sbin/udevd -d
  953 root      25   0  1624  428  372 S    0  0.0   0:00.08 /sbin/mingetty tty1
 1280 root      13  -5     0    0    0 S    0  0.0   0:00.00 [kmpathd/0]
 1281 root      13  -5     0    0    0 S    0  0.0   0:00.00 [kmpathd/1]
 1289 root      13  -5     0    0    0 S    0  0.0   0:00.00 [kmirrord]
 1312 root      10  -5     0    0    0 S    0  0.0   0:01.92 [kjournald]
 1314 root      10  -5     0    0    0 S    0  0.0   3:32.86 [kjournald]
 1316 root      10  -5     0    0    0 S    0  0.0   0:27.39 [kjournald]
 1318 root      10  -5     0    0    0 S    0  0.0   6:53.24 [kjournald]
 1592 phillips  24   0  4620 1504 1224 S    0  0.1   0:00.12 -bash
 1877 root      15   0  1696  576  484 S    0  0.1   1:12.27 syslogd -m 0
 1880 root      18   0  1640  392  328 S    0  0.0   0:00.05 klogd -x
 1897 root      15   0  2268  336  256 S    0  0.0   4:03.23 irqbalance
 1935 rpc       15   0  1776  520  440 S    0  0.1   0:00.00 portmap
 1967 root      15   0  4924  548  280 S    0  0.1   0:00.09 rpc.idmapd
 2001 dbus      15   0  2708  888  704 S    0  0.1   0:00.02 dbus-daemon --system
 2050 root      22   0  6260 1096  856 S    0  0.1   1:03.61 automount
 2074 root      18   0  1644  548  460 S    0  0.1   0:00.00 /usr/sbin/acpid
 2093 root      15   0  5168  944  628 S    0  0.1   0:00.09 /usr/sbin/sshd
 2110 root      18   0  2668  888  716 S    0  0.1   0:00.61 xinetd -stayalive -pidfile /var/run/xinetd.pid
 2160 root      15   0 12068  572  456 S    0  0.1   0:00.87 /sbin/apcupsd -f /etc/apcupsd/apcupsd.conf

gilead · 02-28-2007, 05:01 PM

1.19 is not a very high load average. I haven't had my email server stop sending until the load gets over 12.0. Even then, I've seen figures over 32.0 and the box hasn't crashed - I've just had to manually kill some processes.

Unless you have a very low throughput or errors in your mail logs, I wouldn't be worried by the load averages you've posted.

fukawi2 · 02-28-2007, 05:09 PM

I've managed to get servers to a load average of the low 70's before without crashing :P

I'm not getting errors or anything, but the same server is running apache with SquirrelMail for my users, and with the high L/A apache is getting unresponsive... Perhaps renice'ing apache might help?

I don't really want to give apache priority over mail though to all the lovely people who like to scan the host for vulnerabilities etc...