[Greylist-users] spamd RSET problem - fixed in last release?

[Graham Toal]

> > Given that I have spamd running in front of a live system and am
> > a touch reluctant to experiment with it and break our mail service,
> > can anyone tell me if it should be safe simply to copy the /usr/libexec/spamd
> > binary from a newer OpenBSD system and not make any other changes?
>
> 	Probably not.
> 	You will probably have shared library issues. 

Actually I tried it by killing the libexec process and running from a
copy of spamd in /tmp that I started manually (so if there was a reboot it
would revert to the original code) and all that went wrong was a warning
message about the minor version number of one of the libraries; it still
ran OK.  I've been running out of the copy in /tmp until today and keeping
a very close eye on it.  I really am quite paranoid about making changes
to this system given the number of users - plus rebuilding a new 3.7 system
was just not on the cards due to the hassles of swapping over from one
server to another (plus lack of spare machines).

> You could buy a 3.7 CD :) 	

Yeah, what is the deal with these BSD guys and not putting ISO images
on the net???

> Or get the openbsd src tarball from an openbsd ftp server near you, Or:
> cvs -d anoncvs at anoncvs1.ca.openbsd.org:/cvs get -r OPENBSD_3_7 src/libexec/spamd

That's the ticket!  Worked a treat, thanks for the info.  I've now updated
properly.

> But these are perhaps questions better suited to an OpenBSD mailing 
> list than here, I suggest you take this sort of discussion to 
> misc at openbsd.org or tech at openbsd.org, where more than just I will
> be around to help you :) 

Hint taken :-)  Thanks again.  By the way the level of spam reduction
has been so significant that we're giving serious thought to dropping
our content filtering spam filter and just using the greylisting.  The
content filtering had a very low false positive rate, but any non-zero
false positive rate no matter how low was getting a few negative 
comments from people, and with the spam rate so low now, we can live
with the few spams we get in order to guarantee no false positives
from the Bayesian filter.


And yes, the RSET tweak did solve the issues we'd been having with
that one sender.

G