OT: Re: [Mimedefang] Adding support for learning our addresses

[Jan-Pieter Cornet]

On Tue, Jan 31, 2006 at 05:16:58PM -0600, Sean Ware wrote:
> > They'd have to, or the TCP session would break.
> 
> That's what I was thinking. I was just trying to determine How Evil
> they actually were. Or if some other TCP magic was going on in the
> round-robin. -- At least some small shred of my sanity is retained.

That shred of sanity would quickly wash away once you actually use
one of those devices and tried to trouble shoot problems with it -
especially if loadbalanced boxes are trying to contact another virtual
service that's really serviced by another box but on the same network.
The silent changes to TCP headers are almost impossible to comprehend.

Been there, done that, got the straightjacket.

That said, having a bunch of sendmail/MD/SA boxes behind a loadbalancer
behaves quite good. If one machine accidentally starts eating itself
because some poor schmuck uploaded an mp3 file as .procmailrc, which
procmail always seems to see as an instruction to start forkbombing
and maillooping itself to oblivion, then one box goes down, but nobody
suffers because the machine will be taken out of the pool, and the
service as a whole just continues to run. (Well, you'd have to remove
the erroneous .procmailrc file before this user gets more mail and
takes more boxes down).

-- 
#!perl -wpl # mmfppfmpmmpp mmpffm <pmmppfmfpppppfmmmf at fpffmm4mmmpmfpmf.ppppmf>
$p=3-2*/[^\W\dmpf_]/i;s.[a-z]{$p}.vec($f=join('',$p-1?chr(sub{$_[0]*9+$_[1]*3+
$_[2]}->(map{/p|f/i+/f/i}split//,$&)+97):qw(m p f)[map{((ord$&)%32-1)/$_%3}(9,
3,1)]),5,1)='`'lt$&;$f.eig;                                # Jan-Pieter Cornet