On 3/10/20 5:45 PM, Jeremy Harris via Exim-users wrote:
> I assume you said "multiple" MXs so that you can run system-upgrades
> and planned-maint on them separately, without taking a mail-acceptance
> outage. I'd suggest, given you're wanting that service level, you
> might also want geographically-separate secondary-MX systems for
> backhoe-induced outage and DDOS protection.
>
> I guess another possible reason would be running processing-intensive
> antispam scanning. SpamAssassin used to be notorious for eating cycles,
> but I don't know if that still holds on modern hardware. I don't use it
> myself.
Thanks. Multiple MX's would be for both redundancy and load balancing. I
would either just list them as equal weight MX's or put them behind
HAProxy (or some other load balancer)
There is truth in being able to take one out of rotation and work on it,
etc.
If the NAS goes down, IMAP/POP is going down. While that stinks I would
rather tell my clients to try later rather than explain that we lost
your inbound email.
I understand that if all the MX's go down, remote mail servers should
queue and retry again, but that's not a good way to go about things in
my opinion. I would rather accept it if at all possible and then deliver
it when my systems come back up.
--
inoc.net!rblayzor
XMPP: rblayzor.AT.inoc.net
PGP:
https://pgp.inoc.net/rblayzor/
