On Tue, Jun 29, 2004 at 06:28:26PM -0700, Pat Lashley said:
> I could go with a commandline A/V scanner config in Exim; and use
> something that contacted the remote clamd. Hmm. The clamdscan
> man page doesn't list any parameters to tell it how to contact the
> clamd; but a quick scan of the source seems to show that it gets
> it from the clamd.conf file; so maybe that'll work. Although I'd
> be interested in any statistics on the relative load of running a
> local clamd -vs- the fork/exec/startup overhead of invoking clamdscan
> for each message.
>
> If clamdscan doesn't work with a remote clamd; then I could try
> to fix it; but if I'm going to have to hack any code to do this;
> I'd much rather do a patch for exiscan-acl to make it recur through
> the directory using the clamd STREAM command if the av_scanner
> value points to a non-local IP address.
>
> In the mean time, I suppose I'll have to run clamd locally and
> just hope that it doesn't cause too much load.
clamdscan does not support a remote clamd, but clamav-milter does. I
suppose you could pull some of the code from clamav-milter.c that does
the remote host parsing (passed on the commandline as --server) and add
it to clamdscan. Good luck with it.
FWIW, we're running clamd on several systems that do >75,000 emails/day,
and we're mostly happy with it. Running the milter against a remote
clamd didn't really seem to help anything, and in some cases actually
slowed the machine down.
HTH,
--
--------------------------------------------------------------------------
| Stephen Gran | You never go anywhere without your |
| steve@??? | soul. |
| http://www.lobefin.net/~steve | |
--------------------------------------------------------------------------
