Author: Odhiambo Washington Date: To: exim-users CC: jeff Subject: Re: [Exim] ETRN and disconnected host
* Jeff Green <jeff@???> [20010908 08:16]: writing on the subject '[Exim] ETRN and disconnected host' | I want to get rid of 4 to 8 hour delays in about 40% of incoming email to
| a local office, and believe that proper implementation of ETRN would do it.
| Unfortunately the ISP admin is arguing with me about configuration.
|
| The MTA for the domain is a (dialup) Linux host we'll call m1.mydomain.com
| using Exim 3.12. The following MX records are for the dialup and the ISP's
| queue, respectively:
|
| mydomain.com preference = 10, mail exchanger = m1.mydomain.com
| mydomain.com preference = 100, mail exchanger = queue.isp.net
This is OKAY. I do it this way but read on...
|
| Port 25 is open only to queue.isp.net and internal machines. I'd like the
| ISP to reconfigure to eliminate the delays that result from seeing the
| local computer in either a disconnected state or unwilling to accept
| connections.
|
| The controversial point is whether an MX record is needed for delivery to
| the disconnected host if ETRN is involved. I gather from recent discussions
| that it is not, but the RFC isn't convincing. The ISP mail/DNS admin seems to
| believe that without the first mx record, or reversed priority (above) all mail
| would bounce as there is no local delivery option. It's tough to convince him
| otherwise. I'm gathering fodder to hopefully persuade him, otherwise we'll
| try to find another ISP or hold the connection open during business hours and
| open up port 25 (I'm sorely tempted, but it's not the right solution).
Let me assume that the admin at the ISP is running Exim - because you've posted this
to this list.
The MX config is then okay.
He can set queue_smtp_domains = m1.mydomain.com
That way his server will not try to deliver directly when an e-mail for m1.mydomain.com arrives.
That achieved, he should then configure ETRN for you on his MTA
smtp_etrn_hosts = m1.mydomain.com
Then configure your host to do etrn whenever the ppp link is up! That should be a ppp issue
/etc/ppp/ip-up.local or something.......
| (OT) We have used fetchmail in POP3 mode and all worked well, but somehow
| multidrop mode broke recently and hasn't been addressed, hence the switch to
| ETRN. Our ETRN requests have met with only a '250' response, with no
| indication either during the exchange or by following the mail logs that it's working
| at all.
I don't know why this is happening. Does you host issue
ETRN my.mydomain.com
What does the ISP logs say? Something like
ETRN m1.mydomain.com received from (m1.mydomain.com) [1.2.3.4]
args: /usr/sbin/sendmail -R mydomain.com
Start queue run: pid=14084 -R mydomain.com
That should appear at his end. If not then his server isn't configured properly.
