[exim] strange exim behaviour

Hej,

first off: I know this problem is quite generic, and can be caused by
various things, but I'm asking anyways..

I'm seeing strange "fluctuations" in the message logs. Doing a
# tail -f mainlog|grep Completed
I'm getting sth. like this:

2004-11-25 12:41:25 1CXH2S-0006yW-00 Completed
2004-11-25 12:41:25 1CXHGQ-0007WF-01 Completed
2004-11-25 12:41:25 1CXHZo-0003JB-00 Completed
2004-11-25 12:41:25 1CXHgm-0007Pt-00 Completed
2004-11-25 12:41:25 1CXHjk-0007HC-01 Completed
2004-11-25 12:41:25 1CXH4S-0008Ld-00 Completed
2004-11-25 12:41:25 1CXHGQ-0007Sw-00 Completed
2004-11-25 12:41:25 1CXI05-0003jR-00 Completed
2004-11-25 12:41:25 1CXHgm-0007HC-00 Completed
2004-11-25 12:41:26 1CXHjk-0007Te-03 Completed
2004-11-25 12:41:26 1CXHZo-0003Dz-00 Completed
2004-11-25 12:41:26 1CXHZm-0003GE-00 Completed
2004-11-25 12:41:26 1CXHkk-0007Te-00 Completed
2004-11-25 12:41:26 1CXH4S-0008HI-00 Completed
2004-11-25 12:41:26 1CXHZo-0003Ly-00 Completed
2004-11-25 12:41:26 1CXHZm-0003Hr-00 Completed
2004-11-25 12:41:26 1CXH4S-0008IN-00 Completed
2004-11-25 12:41:39 1CXHjk-0007Te-00 Completed
2004-11-25 12:41:42 1CXHZm-0003NS-00 Completed
2004-11-25 12:41:43 1CXHgk-0007Ml-00 Completed
2004-11-25 12:41:43 1CXHZo-0003Nx-00 Completed
2004-11-25 12:41:44 1CXH4S-0008Kw-01 Completed
2004-11-25 12:41:45 1CXHZk-0003MM-00 Completed
2004-11-25 12:41:45 1CXHZo-0003MM-00 Completed
2004-11-25 12:41:45 1CXH8S-0002b7-00 Completed
2004-11-25 12:41:45 1CXHGQ-0007Wb-00 Completed
2004-11-25 12:41:45 1CXI0M-0003sm-00 Completed
2004-11-25 12:41:47 1CXHWk-0000tw-00 Completed
2004-11-25 12:41:47 1CXH8S-0002QH-00 Completed
2004-11-25 12:41:47 1CXHZo-0003GE-00 Completed
2004-11-25 12:41:47 1CXHGQ-0007TZ-00 Completed
2004-11-25 12:41:47 1CXI0Q-0003wF-00 Completed


So, between 12:41:26 and 12:41:39 no single message got delivered.
Loadaverage is ~15 at that moment, the following settings in exim's
config may be of prior interest:

smtp_load_reserve = 5 # lowered this to 5 during the "test"
queue_only_load = 40
deliver_queue_load_max = 40

There were no messages coming in from remote hosts, exim was processing
mails in its queue. I'm using Berkeley DB files to check local addresses
and their quota on this host. This is on a dual XEON 2.4GHz with 2GB RAM
and SCSI RAID5.

So, can anyone say something like "Ah, I allready know that, it's
because ..."? AFAICS there's no obvious reason for this behaviour. And
tpop3d which is running too on that box, does not reflect this
fluctuations...


lg,
daniel