First, sorry for the long post, but I've tried to include info so we don't have to go back and forth.. :)

courieresmtp has begun crashing on every non-local delivery attempt, logging "crashed child process":

Nov 27 09:13:51 server courierd: started,id=00033A5C.456A2475.00001279,from=<us...@example.com>,module=esmtp,host=hotmail.com,addr=<us...@hotmail.com> Nov 27 09:13:51 server courierd: Waiting. shutdown time=none, wakeup time=none,
queuedelivering=1, inprogress=1 Nov 27 09:13:52 server courieresmtp: Crashed child process 7387, while
delivering to hotmail.com Nov 27 09:13:52 server courierd: completed,id=00033A5C.456A2475.00001279 Nov 27 09:13:52 server courierd: Waiting. shutdown time=Mon Nov 27 09:25:22
2006, wakeup time=Mon Nov 27 09:25:22 2006, queuedelivering=1, inprogress=0

This occurs on courier 0.53.2 from Ubuntu/Debian and 0.53.3 from Debian's testing package repo (which I built myself).

I've setup auth'd relaying to a smarthost, using:

esmtproutes: : [206.190.36.18],587

esmtpauthclient: 206.190.36.18 us...@example.com mypassword

I read last month's discussion of a *very* similar problem, which was caused by the DNS records, and indeed, this is the same ISP and the same server. I want to relay up to "smtp.broadband.rogers.com", which yields this mess:

smtp.broadband.rogers.com is an alias for ssmtp.bloor.is.net.cable.rogers.com. ssmtp.bloor.is.net.cable.rogers.com is an alias for smtp-rog.mail.yahoo.com. smtp-rog.mail.yahoo.com is an alias for smtp-rog.mail.yahoo2.akadns.net. smtp-rog.mail.yahoo2.akadns.net has address 206.190.36.18

I correctly (according to the discussion last month) put "smtp-rog.mail.yahoo2.akadns.net" into both esmtproutes and esmtpauthclient, but when this problem occurred, I changed them (as above) to IP addresses try and determine if it was a name lookup problem. (No change.)

The SMTP conversation starts between courier and the ISP server, courier reads "me", looks for "esmtphelo" (I don't have one), issues an EHLO, the ISP server replies with it's capabilities, courier reads "esmtpauthclient" and then segfaults. (I notice courier is actually talking to "smtp104.rog.mail.re2.yahoo.com", which is 206.190.36.82. But they must be load balancing because on occasion it's smtp103.rog....)

I tried insuring a very straightforward esmtpauthclient file (no comments, spaces not TABs, etc.), in case there is a problem with the file format.

The last 15 lines of strace output:

mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0xb7efe000 read(6, "server.example.com\n", 4096) = 24 close(6) = 0 munmap(0xb7efe000, 4096) = 0 time(NULL) = 1164605267 select(6, [], [5], NULL, {300, 0}) = 1 (out [5], left {300, 0}) write(5, "EHLO server.example.com\r\n", 30) = 30 time(NULL) = 1164605267 select(6, [5], [], NULL, {300, 0}) = 1 (in [5], left {299, 964000}) read(5, "250-smtp104.rog.mail.re2.yahoo.c"..., 512) = 98 open("/etc/courier/esmtpauthclient", O_RDONLY) = 6 fstat64(6, {st_mode=S_IFREG|0600, st_size=48, ...}) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0xb7efe000 read(6, "206.190.36.18 us...@example.com p"..., 4096) = 48 --- SIGSEGV (Segmentation fault) @ 0 (0) ---

I have a tcpdump capture file as well, if it would help.

Courier has delivered messages via this server successfully two or three times in the past. (I'm just setting up courier, so I've only been testing.)

Nov 26 14:34:45 server courieresmtp:
id=0001C8ED.4567F4B6.0000253B,from=<www-data@server>,addr=<us...@hotmail.com>:
250 ok 1164569685 qp 20230 Nov 26 14:34:45 server courieresmtp:
id=0001C8ED.4567F4B6.0000253B,from=<www-data@server>,addr=<us...@hotmail.com>,size=390,success:
delivered: smtp-rog.mail.yahoo2.akadns.net [206.190.36.18] Nov 26 14:34:45 server courieresmtp:
id=0001C8ED.4567F4B6.0000253B,from=<www-data@server>,addr=<us...@hotmail.com>,size=390,status:
success Nov 26 14:34:45 server courierd: completed,id=0001C8ED.4567F4B6.0000253B

Does anyone have any ideas? I'm sure I'm doing something wrong, because and this configuration is quite standard. In fact, I've been running courier for many years, and have done auth'd smarthost relaying in the past.

Although, why is courier segfaulting instead of telling me what's wrong?

Thanks, Richard