lol, I have the same issue again, did a google search and found my old thread.
Is there an answer to this as its bringing the mailserver to its knees each time I sent a mailshot out.
This is an example of the 'top' output and using atop I can see the that box is really disk bound rather than CPU bound, although its a pretty close thing.
Code:
top - 06:19:14 up 28 days, 8:40, 1 user, load average: 7.72, 9.11, 9.63
Tasks: 253 total, 1 running, 252 sleeping, 0 stopped, 0 zombie
Cpu(s): 21.2%us, 1.3%sy, 0.0%ni, 36.2%id, 40.8%wa, 0.2%hi, 0.3%si, 0.0%st
Mem: 3969684k total, 3763208k used, 206476k free, 131120k buffers
Swap: 1048568k total, 116k used, 1048452k free, 651268k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3578 zimbra 15 0 259m 101m 3924 S 16.2 2.6 0:10.77 amavisd
4427 zimbra 15 0 259m 100m 3924 S 15.8 2.6 0:06.93 amavisd
4143 zimbra 16 0 258m 100m 3932 S 3.6 2.6 0:08.71 amavisd
5177 zimbra 16 0 258m 100m 3908 D 3.3 2.6 0:04.39 amavisd
3579 zimbra 15 0 258m 99m 3912 S 1.3 2.6 0:10.90 amavisd
6118 zimbra 15 0 380m 138m 6964 S 1.0 3.6 14:20.51 clamd
4471 zimbra 25 0 557m 73m 7348 S 0.7 1.9 21:43.99 slapd
5868 zimbra 17 0 1579m 1.2g 14m S 0.7 31.4 51:00.32 java
There will be a reasonably high level of bounces, so this will increase the load of course - these should drop over time when the code in the mail system removes them from the mailing list.