Sorry for reposting this (I sent it to fedora-list earlier), but I'm running out of ideas...
I'm having a weird problem... I'm trying to set up a mail server for a client. It's a dual Opteron box with 4GB RAM running FC2/x86_64.
I'm setting it up with amavisd-new, which requires setting up postfix to listen on a port (so postfix relays to amavisd which relays to postfix). The initial connection to postfix (port 25) is fine. The mail is processed by amavisd fine. When amavisd tries to connect to postfix again, I get this:
==> /var/log/messages <== Aug 31 23:33:20 foo kernel: smtpd[1665]: segfault at 00000030bf300e20 rip 00000030bf206294 rsp 0000007fbfffd8b0 error 7
==> /var/log/maillog <== Aug 31 23:33:20 foo postfix/master[1394]: warning: process /usr/libexec/postfix/smtpd pid 1665 killed by signal 11 Aug 31 23:33:20 foo postfix/master[1394]: warning: /usr/libexec/postfix/smtpd: bad command startup -- throttling
I've tried upgrading and downgrading the kernel to every FC2 errata kernel-smp, plus I've tried kernel-smp-2.6.8-1.533 from rawhide. I've also tried upgrading postfix to 2.1.4-1 from rawhide. Nothing seems to change the behavior.
It's entirely possible that I've screwed something up somewhere that explains all this, but I'm inclined to think not... I've basically cloned a working config from another (i386) box.
Any ideas would be very welcome...
Steve
On Wed, 1 Sep 2004, Steven Pritchard wrote:
I'm setting it up with amavisd-new, which requires setting up postfix to listen on a port (so postfix relays to amavisd which relays to postfix). The initial connection to postfix (port 25) is fine. The mail is processed by amavisd fine. When amavisd tries to connect to postfix again, I get this:
==> /var/log/messages <== Aug 31 23:33:20 foo kernel: smtpd[1665]: segfault at 00000030bf300e20 rip 00000030bf206294 rsp 0000007fbfffd8b0 error 7
==> /var/log/maillog <== Aug 31 23:33:20 foo postfix/master[1394]: warning: process /usr/libexec/postfix/smtpd pid 1665 killed by signal 11 Aug 31 23:33:20 foo postfix/master[1394]: warning: /usr/libexec/postfix/smtpd: bad command startup -- throttling
Does this happen every time, or sporadically?
Back out the IPv6 and TLS / AUTH patches and see if you can reproduce it.
If you can, follow the steps in /usr/share/doc/postfix-mumble/README-FILES/DEBUG_README (Wietse won't support either patch until he merges them for 2.2, so don't waste his time sending in reports produced with them applied)
I've been having the same sorts of sporadic stability issues when I apply the IPv6 patch on recentish RHEL / Fedora boxes with Postfix 2.1.x.
later, chris
On Wed, Sep 01, 2004 at 04:43:54PM -0400, Chris Ricker wrote:
On Wed, 1 Sep 2004, Steven Pritchard wrote:
I'm setting it up with amavisd-new, which requires setting up postfix to listen on a port (so postfix relays to amavisd which relays to postfix). The initial connection to postfix (port 25) is fine. The mail is processed by amavisd fine. When amavisd tries to connect to postfix again, I get this:
==> /var/log/messages <== Aug 31 23:33:20 foo kernel: smtpd[1665]: segfault at 00000030bf300e20 rip 00000030bf206294 rsp 0000007fbfffd8b0 error 7
==> /var/log/maillog <== Aug 31 23:33:20 foo postfix/master[1394]: warning: process /usr/libexec/postfix/smtpd pid 1665 killed by signal 11 Aug 31 23:33:20 foo postfix/master[1394]: warning: /usr/libexec/postfix/smtpd: bad command startup -- throttling
Does this happen every time, or sporadically?
Every time.
I tried un-installing postfix and installing the i386 postfix from rawhide... With the exact same configuration, the i386 postfix works perfectly, so there's definitely a bug here somewhere...
Steve