[Mimedefang] A big recommendation for using ramdisk on MDspooldirectory

[Kris Deugau]

Richard Laager wrote:
> We've had our /var/spool/MIMEDefang directory on a tmpfs filesystem
> for quite a while. Every once and a while, mail stops flowing because
> the tmpfs filesystem is full. That by itself is strange. It's even
> stranger because when I rm -rf /var/spool/MIMEDefang/* (after
> stopping MIMEDefang and Sendmail), the disk usage stays the same. I
> even made sure that all the mimedefang* and sendmail processes were
> killed.

How big is your tmfs mount?

How much physical memory do you have?

I've had no trouble other than a CPU bottleneck on the box I have using
tmpfs on the MD spool.

> The box is running Red Hat Linux 9. The kernel version is 2.4.20-18.9
> from the RedHat RPM.

I *think* a newer RH kernel should be available;  there has been a
bugfix release and a security release for 7.x/8.0 within the past month.

> Has anyone seen this sort of behavior?

Not on the one server I have using tmpfs;  it's RH7.3/kernel 2.4.20-24.7
(which I should really reboot for the recently released security fix)
with a 96M tmpfs mount in 512M physical memory.  Uptime has been more or
less continuous for almost a year in this configuration aside from
kernel-replacement or need-to-rearrange-UPS-load downtime.

-kgd
-- 
"Sendmail administration is not black magic.  There are legitimate
technical reasons why it requires the sacrificing of a live chicken."
   - Unknown