I have read the FAQ and the Majordomo manual, and I couldn't find any reference to this
particular problem. I appreciate any area someone can point to that I can look at for this.
I am installing Majordomo 1.93 on two different Linux boxes (1.2.1) and I get the same
results on both. Perl is 4.036. I set up the aliases, config file, directory structure,
etc. per Jerry Peek's chapter in O'Reilly's MIIS. Permissions are good (files 664, dirs
775) as far as I can see, and all the files are owned by majordom.majordom.
I found two problems, supposedly in the 1.92 release, were still in the 1.93 that I got
(from CERT, I think); the line numbers were different, but the described problem is the
same. The 'resend' file (line 265) had an unlink("/tmp.....") which I changed to
unlink(</tmp...>), and the 'request-answer' file (line 41) sent mail to "-f$list-request"
instead of "-f$list-approval". Neither of these changes had any effect.
Aliases are:
owner-owner: root
nobody: /dev/null
majordomo: "|/usr/local/majordomo/wrapper majordomo"
owner-majordomo: root
majordomo-owner: owner-majordomo
(There are a couple of others on the box at home that I can't remember. The above is from
the docs that come with Majordomo 1.93.)
When I send mail to majordomo@myhost.com with 'lists' in the body, perl is started running
the wrapper. I then see two sendmails spawned in the order listed:
sendmail -oi -oem -t
sendmail -fMajordomo-Owner@myhost.com -t
Shortly after that, the perl process takes off on a wild memory gobble, grabbing all the
memory it can get, forcing everything to swap. Watching memory use, it is heading towared
exhausting my swap partition; I don't let it get that far. When I kill -9 the perl process
everything settles down and the mail reply for 'lists' is returned to the posting user as if
nothing had gone wrong.
There are no tracks in the logs except for a message that sendmail died with a signal 11.
Sendmail is 8.6.11, and it has no history at all of failure on either box. (Related to my
killing it with signal 9??? Uncertain.)
If I don't kill pending mail in /var/spool/mqueue, this all starts over again when sendmail
wakes up and tries to send queued mail.
As I indicated, I found no reference to such a problem in the FAQ or manual; I tried to do
my homework up front, but nothing has helped so far. Any help is appreciated, and thanks in
advance.
Michael Brennen
|
|
