Summary: sendmail problems

From: Jim Faust (jbf@pixel.cirrus.com)
Date: Thu Apr 03 1997 - 08:18:26 CST


Original Inquiry:

>when I came in this morning I found that no mail had been delivered since
>about six last night. A quick look at /var/spool/mqueue showed that all mail
>was undelivered and still pending. A look in /var/adm/messages showed the
>following:
>
>Apr 2 08:55:53 homebrew sendmail[702]: AA00584: SYSERR: putbody: read error
>Apr 2 08:55:53 homebrew sendmail[702]: AA00584: SYSERR: mail died because of
>IOT instruction (6)--requeueing message
>Apr 2 08:55:54 homebrew sendmail[702]: AA00522: SYSERR: mail died because of
>IOT instruction (6)--requeueing message
>Apr 2 08:58:31 homebrew sendmail[729]: AA00727: SYSERR: putbody: read error
>Apr 2 08:58:31 homebrew sendmail[729]: AA00727: SYSERR: mail died because of
>unknown error code (134)--requeueing message
>Apr 2 08:58:57 homebrew sendmail[740]: AA00738: SYSERR: putbody: read error
>Apr 2 08:58:57 homebrew sendmail[740]: AA00738: SYSERR: mail died because of
>unknown error code (134)--requeueing message
>Apr 2 08:59:00 homebrew sendmail[744]: AA00735: SYSERR: putbody: read error
>Apr 2 08:59:00 homebrew sendmail[744]: AA00735: SYSERR: mail died because of
>unknown error code (134)--requeueing message
>
>attempts to force delivery with sendmail -Rusername resulted in a core dump,
>as do many sendmail commands.
>
>I've replaced sendmail with a known good binary, although doing a check sum
>showed that it wasn't corrupted, and I also restarted both sendmail and the
>entire system.
>
>The system in question is a Sparc 2 running SunOS 4.1.2. Nothing has been
>changed on this system for months, and it's been running sendmail for at
>least three years without a problem.
>
>Any ideas?
>

Something else that I didn't mention was that when attempting to flush
the queue using "sendmail -Rusername", along with the core dump, I received
the following error:
 
  assertion botched: op->ov_magic == MAGIC

Solution:

I performed a wide range of checks and even got into dbx and trace.
Karl Vogel gave me a really neat command to run diags with:
echo test | /usr/lib/sendmail -d0-99.127 djp@inetinc.com > /tmp/sendmail.dmp
It didn't help but I'll keep it for later use.

Two people suggested that I had a malformed mail message in the queue. I
looked into this early on and moved pending messages out of the queue. The
mail somewhat worked for a while after cleaning things out but quickly
returned to the above errors.

Another suggestion was to look into my ISP, but since I have a direct
connection I was the only one to blame for my problem.

Ultimately the problem was me, I had modified my /etc/group file a day
earlier and inadvertently left a newline at the base of the file.

Thanks to:

djp@cliff.inetinc.com David J. Pagliai
vogelke@c17.wpafb.af.mil Karl E. Vogel
jharmon@telecnnct.com Jim Harmon
bismark@alta.Jpl.Nasa.Gov Bismark Espinoza
burelbac@ncifcrf.gov Jonathan Burelbach



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:49 CDT