My original question was:
What happens to incoming mail if the destination machine is taken
off-line for a few days?
My thoughts are that the mail delivery will be tried for 3 days out on
the Internet and if the host is not up by then, the sender will receive
a notification that the mail cannot be delivered- since this has
happened to me on occassion with some mail I have sent.
If this is true, is this universal behavior?
Summarizing the responses-
The amount of time that the sending machine will try to deliver is
configurable- must default to 3 days under Solaris since I did not
change anything related to this in my mail config. One suggestion was
to establish another mailhost outside my domain.
The replies follow:
************************************
>From george@pop.psu.edu Fri Jun 21 09:39 EDT 1996
I think yes!!! (although it depends on how are the sende's site configured
(i.e. they may try for (say) 5 days).
You should probably start using MX records to have an alternative mail host
(even one outside csph).
************************************
>From robbie@tomservo.mindspring.com Fri Jun 21 09:45 EDT 1996
It all depends on how each mailhost has their sendmail set up.
Ours here are set to try to send mail for 3 days, then give up.
I've seen others that give up after one day. And a few others
that are set to 5 days.
Of course, if you're set to try to send a message for 5 days,
that means that your server has to queue the message for five
days, and keep trying to send it. THis can bog down a machine
if it has a large volume of mail waiting to be sent.
There is no universal behavior- mail daemons are configured
differently all around the world. What matters to you, though,
is what the server that you're using is set to.
************************************
>From ca@mine.informatik.uni-kiel.de Fri Jun 21 09:56 EDT 1996
This is configurable.
Our mailserver tries 4 days, so it depends on the configuration of
the sender.
************************************
>From sagray@amp.com Fri Jun 21 10:01 EDT 1996
Hi, Craig. I believe that this is part of the sending
host's configuration, and it's not universal. It's
probably a good rule of thumb, though.
If you're expecting significant downtime for a machine
that recieves a lot of mail, consider adding an alias
on your mail-forwarding service or on your naming service
(be it DNS, NIS, or NIS+) to point, at least temporarily,
to another machine.
************************************
>From sahirns@menger.eecs.stevens-tech.edu Fri Jun 21 10:03 EDT 1996
If your machine cannot deliver the mail to the destination machine, it
keeps trying. It periodically sends warning messages to the sender
notifying them that the message has not yet been delivered. After 3-4
days (4 on my machine) the mail bounces back to the sender.
-s
************************************
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:02 CDT