[Mimedefang] Re: "Please try again later"
John Rudd
john at rudd.cc
Sun Sep 11 02:38:06 EDT 2005
On Sep 10, 2005, at 7:40 PM, David F. Skoll wrote:
> By the way, did you make sure that /var/spool/MIMEDefang was on a
> tmpfs volume on your Sun machines? If you don't do that, it will kill
> performance. You might want to verify that.
>
The load is very different right now, but that certainly seems to be
making a difference. I hadn't even considered that (with the
dual-queue approach in mailscanner, you kind of don't want to do that,
because if the computer crashes, you may very well lose messages).
So, I'm running one of the 4 production machines that way right now.
I'll wait and see how it does tomorrow before I decide what to do next.
Very cool.
(I also added to the "start)" and "stop)" parts of the rc script, so
that it mounts and unmounts that tmpfs when you start and stop
mimedefang ... and it also has to start and stop clamd in there, as
well, since its socket is in that directory)
Thanks very much for the feedback (on a Saturday no less).
More information about the MIMEDefang
mailing list