This is a good and helpful answer, but I'm going to argue with your
contention that
The current behaviour is absolutely right.
I presume that fetchmail is designed with the following three goals in
mind:
1. To deliver mail without losing it;
2. To deliver each item of mail once only;
3. To deliver mail in a timely manner.
Obviously when POP3 is the source of mail and one or more e-mails
cannot be delivered, there is a problem, and the current
implementation of fetchmail "resolves" this by abandoning goals 2&3.
If you receive multiple copies of message or others are delayed (not
fetched) because of a mail with a temporary error, this would be a bug
(I think) and has to be fixed. A complete log would be helpful in this
case.
I think that something else is going on, and that my original analysis was
missing something. I tried your suggestion of the sendmail
`accept_unresolveable_domains' feature, and it makes little difference.
Instead, looking at the log file, it seems that the POP3 connection to my
ISP may be timing out while my sendmail is trying to perform the domain
lookup.
This make sense, and sounds rather tricky to resolve. I don't know
what the timeout on a POP3 connection should be, but it certainly looks as
if my ISP isn't going to give my fetchmail time to do anything time
consuming with the retrieved mail.
Here is the last entry in the brief log, with unresolveable domains
enabled:
fetchmail: 5 messages for me at pop.ntlworld.com (18691 octets).
fetchmail: reading message me(_at_)pop(_dot_)ntlworld(_dot_)com:1 of 5 (2336
octets) fetchmail: flushed
fetchmail: socket error while fetching from pop.ntlworld.com
fetchmail: Query status=2 (SOCKET)
fetchmail: sleeping at Wed Nov 20 10:27:34 2002
Notice the socket error. This is, I believe, due to the connection being
closed by pop.ntlworld.com, presumably due to the long delay in failing to
perform the domain lookup on mail1.sarahsnaughtymail.com.
An extract from a full log (sendmail -v -v) seems to confirm that the
connection to pop.ntlworld.com has been dropped:
...
fetchmail: forwarding to localhost
fetchmail: SMTP> MAIL
FROM:<bounce-ns1-846773(_at_)mail1(_dot_)sarahsnaughtymail(_dot_)com> BODY=7BIT
SIZE=2336
fetchmail: SMTP< 250 2.1.0
<bounce-ns1-846773(_at_)mail1(_dot_)sarahsnaughtymail(_dot_)com>... Sender ok
fetchmail: SMTP> RCPT TO:<guest(_at_)localhost>
fetchmail: SMTP< 250 2.1.5 <guest(_at_)localhost>... Recipient ok
fetchmail: SMTP> DATA
fetchmail: SMTP< 354 Enter mail, end with "." on a line by itself
#fetchmail: SMTP>. (EOM)
fetchmail: SMTP< 250 2.0.0 gAKAdB3x014087 Message accepted for delivery
fetchmail: flushed
fetchmail: POP3> DELE 1
fetchmail: 6.1.0 querying pop.ntlworld.com (protocol POP3) at Wed Nov 20
10:43:12 2002: poll completed
fetchmail: Query status=2 (SOCKET)
fetchmail: Deleting fetchids file.
fetchmail: normal termination, status 2
fetchmail: Deleting fetchids file.
Sunil Shetye has suggested I upgrade from my current version, 6.1.0, to
6.1.2. As my fetchmail is built from FreeBSD ports and as the latest
version in ports is only 6.1.1, and as 6.1.0 is described on the fetchmail
web site as the "gold" version, unless this problem has been specifically
addressed in 6.1.2 I'd rather not attempt the upgrade.