70k mqueue.in but load under 1 ??

[Tony Johansson]

I've reinstalled our MailScanner servers with CentOS4.4 (previous 
centos3.x) and upgraded memory from 1 to 2Gb.
Servers are HP Proliant DL380G3 running at 3.0Ghz

They process about 80k messages/day each with round-robin dns records.

On occasion the queues build up (this morning it was 70k!) but the load 
on the machines go way down, usually under 1.

"sar" shows:
10:00:01 AM       CPU     %user     %nice   %system   %iowait     %idle
10:10:02 AM       all     49.72      0.00     10.70      1.60     37.98
10:20:01 AM       all     45.30      0.00     10.20      1.54     42.96
10:30:02 AM       all     47.36      0.00     10.24      1.58     40.82

MailScanner version 4.56.8
5 children 30 messages
SpamAssassin 3.1.7 with use_bayes 0

time spamassassin --lint -D
real    0m2.596s
user    0m1.345s
sys     0m0.097s

rules_du_jour not active
clamav and bitdefender
local caching dns (named)
2.6.9-42.EL kernel

bonnie++ looks fine?:
ms01 2G 24176 77 52571 24 20910 7 25757 69 57056 8 558.2 1 16 +++++ +++ 
+++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++

Used to run smoothly until the upgrade.

Any ideas on whats going on here and what I can do to troubleshoot?

Regards, Tony