[Pkg-exim4-users] Admin'd exim for a long time under RH, exim question about queue_only & configuring Debian's exim4 package in general

[Josh Matthews]

Greetings all, and first let me say congratulations to all those involved in 
the Debian project in general, and specifically I'd like to thank those 
responsible for doing such a versatile and powerful packaging job with 
Exim4. Sarge is my first experience with Debian on one of my servers, and as 
they say - "it just keeps getting better". I cut my teeth on RH 7.0 and ran 
it through 9, but after being left "high and dry" decided not to go down the 
Fedora road. Necessity dictates a major server overhaul, and Debian's won 
both my respect and admiration due to it's structure, capabilities, and 
community. 

Now for the task at hand: our main mail server is located in our public 
offices, however we've recently signed a colocation agreement with a 
provider about 3 hours from our location. All of our other systems (Web 
hosting & SQL, DSL provisioning equipment, etc) has been relocated (with the 
exception of DNS), however now we find it's time to move the mail server. 
Since it's a physical move, we can't avoid turning off the power for 2 or 
three hours to server which is the lowest numbered MX record for all of our 
domains, plus the domains we host. Obviously this is something that can be a 
little distressing.

My plan has been to create a secondary server, which will temporarily take 
over the primarys server's name in DNS. It's that new server which has been 
configured with Sarge's exim4 distribution. I intend to set that server to 
"queue_only" while the second server is being transported & installed in the 
co-lo. Once it's been brought up on the new IP range, I intend to change DNS 
once again for both our internal & hosted domains, so that our primary MX 
record is the (now) colocated server, and change my backup server's DNS to 
be the MX 20 of those domains. From what I can gather, once I initiate a 
queue runner process on the backup server, it will treat all queued 
"internal" mail just like all the other mail - i.e. using remote_smtp. My 
question is two-fold:

1) Does this like a reasonable solution, to those more experienced with 
Exim? About the only other conclusion I could come up with was to allow the 
backup server to deliver mail normally - local addresses to mbox spools & 
remote_smtp for everything else. Than just append the contents of each old 
mbox spool from the backup server to the user's spool on the new server.

2) Considering that regardless of the configuration solution I use to 
achieve this, I'll be keeping this server with a function Exim4, what's the 
recommended method of creating the (extensive) macro file I'll need? My best 
plan for that so far was to go through each of the conf.d/ files with a 
yellow notepad and a pen, make a note of the dc_ variables and the values I 
intend to change them to. Is this about what everyone does?

Thanks much for taking the time to read (and hopefully reply) to this, and 
again special thanks to the maintainers of the exim4 package. No matter 
what, I'm with this "for the long haul" :)

Josh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.alioth.debian.org/pipermail/pkg-exim4-users/attachments/20050807/e407d74a/attachment.htm