[Mimedefang] My MD install went wacko

[Justin Shore]

On Sun, 8 Jun 2003, Jeffrey Goldberg wrote:

> On Sat, 7 Jun 2003, Justin Shore wrote:
> 
> > During the temp failing period the remote MX queued up many thousand
> > messages for my spamtrap. [...] A quick restart of MD and Sendmail (one
> > or the other wouldn't fix it.  I had to do both) fixed the problem.
> > The fix was only temporary though as it happened again a short time
> > later.
> 
> > This is a RH 7.3 box running Sendmail 8.12.9, [...]
> 
> Do you have a sendmail submission (client) queue runner daemon working?
> 
> It sounds like you don't.  Since 8.12.9, sendmail splits its functions (to
> avoid running suid) and you need to have a client daemon running using
> submit.cf if there is going to be any locally generated mail.

I have it running.  Everything Sendmail seems to running fine.  I have to 
restart it only to make it check the milter again sooner rather than 
waiting for it to get around to it.  IIRC Sendmail won't check a failing 
milter for X amount of time after the first failure or two.  Restarting 
Sendmail kicks it in the butt and speeds things up.

> I'm not sure that this is part of the problem.  It certainly doesn't
> explain the initial MD errors.  Can you post your .mc file declaration of
> the mimedefang Milter.  There might be a way to tune it so sendmail fails
> more gracefully if it can't talk to MD.

I suspect the problem lies within MD but I haven't got a clue where.  I 
just upgraded to beta5.  It detects CLAMSCAN, CLAMD, and FPROT at 
configure time.  I'm assuming it calls both clamscan and clamd.  If so 
then that's redundant.  I'm not sure which I should disable though.  IIRC 
you're supposed to use clamdscan (note the d) when using the clamd daemon, 
which I am.  I'm not sure this matters though with regards to this 
problem.

Here's the relevant part of my sendmail.mc.

define(`MILTER', 1)
dnl MIMEDefang and SpamAssassin
INPUT_MAIL_FILTER(`mimedefang', 
`S=unix:/var/mail/MIMEDefang/mimedefang.sock, F=T, T=S:60s;R:60s;E:5m')

The load on the machine is usually negligible.  I run two instances of the 
Folding @ Home client on this machine which brings the load up to 2 (it's 
niced down to background noise).  The Highpoint driver for my ATA RAID 
card adds 1 to the load as well.  Don't ask me why cause I don't know.  
The workload on the CPUs doesn't really go up because of the drivers 
but for some reason it reports it as such.  Nevertheless I don't think 
load is an issue.

If anyone wants large amounts of logs or my mimedefang-milter drop my a 
line and I'll send it off list.  I don't want to bog down the list with 
it.

Thanks
 Justin