just cvsup'ed to what i thought was -stable (RELENG_4), built and installed
world and kernel 12 hours back. uname -a shows FreeBSD 4.3-BETA though.
mergemastered everything in. all works fine, except this odd quirk with sendmail
i've got sendmail configured (did it new with the upgrade) for
DeliveryMode=deferred and a smart host since i'm on a dialup. when ppp sings
it's mating song and connects, i run 'sendmail -q -v' to clear the queue. i
notice that it seemingly hangs for a long time (50 seconds) before pushing the
whip out ethereal and snoop the ppp0 interface, and here's where i observe this
weird behaviour. i see a flurry of DNS requests for the AAAA (IPv6) record of
the smarthost, but goes out as a IPv4 packet. finally (after 1.5 seconds), an A
record query for the smarthost goes out and is promptly answered by the
then here's the odd part, sendmail attempts to connect (SYN) to port 25 of the
smarthost, but the TTL of the outgoing tcp packet is set to 0. yes, that's zero.
predictably, the next hop returns an ICMP TTL Exceeded packet. this continues
for another 58 seconds and 9 packets. finally, sendmail tries again (on a new
local source port, so it seems to be a new connection) and this time setting the
TTL to the correct value of 64. this connection goes thru and mail gets sent.
problem only exists with sendmail, as telnet, nntp, pop3 and a host of other tcp
proggies work fine. ipv6_enable="NO" in rc.conf and sendmail's DAEMON_OPTIONS
are the default without the addition of IPv6.
i suspect it's a sendmail issue, as nothing else is affected and works the way
it's supposed to. any clues as to what's causing this, as the delay is
irritating and causes me to stay dialled in longer than i really need to.
will be more than happy to send the ethereal libcap saved file to anyone
Regards, /\_/\ "All dogs go to heaven."
| for a in past present future; do |
| for b in clients employers associates relatives neighbours pets; do |
| echo "The opinions here in no way reflect the opinions of my $a $b." |
| done; done |