"Stalled" tcpsvd?

Hi,

I am experiencing a strange symptoms on a couple of my systems, where
tcpsvd would in a sense "stall", thinking that all connections are coming
from the same IP address.

Here is a excerpt my the logs where is starts:

2004-11-26_19:04:51.98418 tcpsvd: info: status 0/30
2004-11-26_19:04:55.85546 tcpsvd: info: status 1/30
2004-11-26_19:04:55.85722 tcpsvd: info: pid 5304 from 211.185.38.61
2004-11-26_19:04:55.85938 tcpsvd: info: start 5304 tirith.elendil.ru:195.68.151.
124 ::211.185.38.61:3619 ./peers/0
2004-11-26_19:04:55.99484 rblsmtpd: 211.185.38.61 pid 5304: 553 Spam Received Se
e: http://www.dnsbl.sorbs.net/lookup.shtml?211.185.38.61
2004-11-26_19:04:56.43241 tcpsvd: info: end 5304 exit 0
2004-11-26_19:04:56.43303 tcpsvd: info: status 0/30
2004-11-26_19:05:03.21712 tcpsvd: warning: unable to accept connection: connecti
on aborted
2004-11-26_19:05:04.86597 tcpsvd: info: status 1/30
2004-11-26_19:05:04.86789 tcpsvd: info: pid 5308 from 211.185.38.61
2004-11-26_19:05:04.86957 tcpsvd: info: start 5308 tirith.elendil.ru:211.185.38.
61 ::211.185.38.61:3619 ./peers/0
2004-11-26_19:05:04.88147 rblsmtpd: 211.185.38.61 pid 5308: 553 Spam Received Se
e: http://www.dnsbl.sorbs.net/lookup.shtml?211.185.38.61
2004-11-26_19:05:05.75259 tcpsvd: info: end 5308 exit 0
2004-11-26_19:05:05.75322 tcpsvd: info: status 0/30
2

After the "connection aborted" message, all SMTP connections are coming 
from 211.185.38.61. This situation continues from 2004-11-26 19:04:55 GMT
to 2004-11-29 08:37:44 GMT (almost 2 days, the whole weekend), until I noticed
this myself and did "runsvctrl t /service/smtp":

2004-11-29_08:37:44.12964 tcpsvd: info: status 0/30
2004-11-29_08:38:20.30458 tcpsvd: info: sigterm received, exit.
2004-11-29_08:38:20.35341 tcpsvd: info: listening on 0.0.0.0:smtp, uid 82, gid 8
1, starting.
2004-11-29_08:38:49.91016 tcpsvd: info: status 1/30
2004-11-29_08:38:49.91226 tcpsvd: info: pid 15602 from 216.136.204.119
2004-11-29_08:38:49.91413 tcpsvd: info: start 15602 tirith.elendil.ru:195.68.151
.124 ::216.136.204.119:23730 ./peers/0

Here is my /service/smtp/run:

#!/bin/sh
exec 2>&1
exec env - PATH=/pkg/bin:/pkg/sbin \
tcpsvd -v -u qmaild -l `/bin/hostname` -i ./peers 0 smtp \
rblsmtpd -b \
        -r dul.ru \
        -r dnsbl.sorbs.net \
        -r combined.njabl.org \
        -r relays.ordb.org \
        -r list.dsbl.org \
        -r bl.spamcop.net \
        -r sbl-xbl.spamhaus.org \
smtpfront-qmail

For what it worth, I have experienced the same symptoms on this system (my
company's server), as well as mail server for one of my clients.
It happened few times already, and each time I had to manually restart
tcpsvd. I have originally attributed this behaviour to extremely high load
on my client's server, but my company's server has very light load.

The systems have different hardware, but similar software config:

FreeBSD 5.2.1-RELEASE
ipsvd-0.9.6

Any ideas on what might be causing this behaviour?

Thanks,

Sergei

-- 
Sergei Kolobov, FreeBSD committer <[email protected]>
GPG key fingerprint: A2F4 5F34 0586 CC9C 493A  347C 14EC 6E69 3BA5 3401

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (FreeBSD)

iD8DBQFBqwplFOxuaTulNAERAhP4AJ0Q8DdnZkUtf/1bkROm8a4XApR2HQCeNjp3
N8E4d/qpsyPqM4Q8G3LrS6c=
=oh3R
-----END PGP SIGNATURE-----