On 2012-10-05 10:40, Chris Green wrote:

As per the subject line, do most SMTP servers (as in systems 'out there'
which send me mail) behave reasonably sensibly when my SMTP server is
temporarily not functioning?

I currently have both a desktop machine *and* a small server machine
which are on all the time.  It's the server machine that receives my
E-Mail. The only reason I have it configured this way is because I'm
more likely to rebuild/reconfigure my desktop machine than the server
machine and I felt that keeping my SMTP server online consistently was
important. 

However I'm wondering if the effort (and cost) of running two systems
all the time is worth it.  The other functions of the server system (DNS
and CUPS) can easily be run on my desktop machine, occasinal downtime of
those services is easily explained to other family members! :-)

If real world SMTP servers 'out there' will cope quite happily with my
domain's MX machine being AWOL for several hours then I may as well put
everything on the one server machine.

(Note it's only *my* E-Mail that is delivered by SMTP, all other family
members do their own thing as in use Webmail, or IMAP, or whatever)

As I understand it, mail servers will retry with decreasing frequency for a set amount of time (dependent on the settings on the server in question - usually up to about 7 days where the destination host is unreachable or unresponsive) - senders may see undeliverable messages stating that the server will keep trying for a further x days/hours before giving up.

The RFC (http://www.rfc-editor.org/rfc/rfc5321.txt) in section 4.5.4 specifies the standards for retrying sent messages - a client must delay after failing to deliver mail to a receiving server, and it makes suggestions as to the frequency of these retries and the delay before bouncing them.

HTH,
Jim