[Mimedefang] Please try again later errors

[Jason Gurtz]

On 10/14/2004 15:23, Trevor Dodds wrote:
> Hi,
> 
> I'm running 2 MimeDefang 2.45 with SA 3 on a Sun Fire V210 Servers both with
> MX 10 with 1 GB ram.
> My mimeconfig is as follows:
> /usr/local/bin/mimedefang-multiplexor -U defang -p
> /var/run/mimedefang-multiplexor.pid -m 8 -x 20 -b 600 -l -r 200
> Everything seems to work fine until someone sends out a mailing list and I
> have thousands of emails being
> Processed,  I run out of memory on the server as mimedefang.pl is using
> around 50MB + 50MB per process X 20.
> Then I receive lots of tempfails followed by please try again later.
[...]
> Will adding another 1GB of memory and another CPU help?  

I would cut down max slaves amount some to maybe 10-15 or so, or add
memory.  I guess it depends on what else is running on the box but I shoot
for about 50-75% of available RAM for the slaves.

On vastly slower hardware I'm using:

define(`confMAX_DAEMON_CHILDREN',     `30')dnl
define(`confCONNECTION_RATE_THROTTLE',`2')dnl
define(`confDELAY_LA',                `2.00')dnl
define(`confREFUSE_LA',               `15.00')dnl

$MULTIPLEX -x 6 -m 2 -q 30 -Q 120 -r 500 -b 300 -i 600 -U defang -E -l

I think you should try the embedded interpreter (-E) if you can, as I saw
a huge performance boost from that.  Using the multiplexor queue feature
also helps some but not as much, by far, as the embedded interpreter.
Also, since most of the CPU is taken by spam assassin, you may want to
look at *not* scanning mail from local sources.

I'm not very sure of the sanity of my multiplexor queue settings or
timeouts but I haven't thoroughly benchmarked so...

Cheers,

~Jason

--