Great Circle Associates Majordomo-Users
(March 1997)
 

Indexed By Date: [Previous] [Next] Indexed By Thread: [Previous] [Next]

Subject: Re: shlock problem
From: Dave Wolfe <dwolfe @ risc . sps . mot . com>
Date: Mon, 17 Mar 1997 09:46:59 -0600 (CST)
To: byrnejb @ harte-lyne . ca (James B. Byrne)
Cc: majordomo-users @ greatcircle . com (Majordomo user's mailing list)
In-reply-to: <199703171509.KAA26342@hahp9k.harte-lyne.ca> from "James B. Byrne" at Mar 17, 97 10:03:54 am
Reply-to: Dave Wolfe <david_wolfe @ risc . sps . mot . com>

[ majordomo-workers dropped from Cc list ]

[ James B. Byrne writes: ]
> 
> I began receiving these messages over the weekend.  Can anyone tell 
> me what they mean? What is causing them? How do I fix the problem?
> 
> MAJORDOMO WARNING (mj_majordomo)!!
> 
> shlock failed: write('/var/spool/majordomo/logs/shlock.25275',
> '25275'): Bad file number
> 
> The contents of the majordomo.debug file are:
> 
> mj_majordomo: WARNING
> shlock: open('/var/spool/majordomo/logs/L.majordomo.log'): No such
> file or directory mj_majordomo: WARNING shlock failed:
> write('/var/spool/majordomo/logs/shlock.25275', '25275'): Bad file
> number mj_majordomo: WARNING shlock:
> unlink('/var/spool/majordomo/logs/shlock.25275'): No such file or
> directory mj_majordomo: WARNING shlock:
> unlink('/var/spool/majordomo/logs/L.majordomo.log'): No such file or
> directory

[ Owner/permissions on log file and directory deleted but look OK ]

Check to be sure that you haven't (temporarily) run out of disk or
reached a quota on that disk volume. You don't say what version
of Mj you're running, but it looks like a vintage 1.94. There's
been some cleanup in shlock.pl since then to abort hopeless
situations and not whine about trivial irregularities. I hope that
/var/spool/majordomo/logs isn't an NFS mount, but that's the only time
I've seen "Bad file number" messages: the open works but the system file
descriptor is no good when the session subsequently fails. Mj won't work
on NFS mounts because file locking fails. If you absolutely *must* use
NFS, I published an flock() version of shlock.pl on Mj-workers recently
that works for me.

> Plus I am getting a bunch of log entries from STAPLES.COM that look 
> like this:

[ Multiple 'help' log entries deleted ]

> Now it looks to my untutored eye that someone is either deliberately
> or accidentally banging Mj to death by repeated help requests.

More likely is a site that's replying to majordomo messages. When Mj
doesn't recognize commands it triggers a 'help' response that gets
logged just like an actual 'help' request. If you aren't archiving Mj
requests, it's hard to tell exactly what's going on, but you might need
an additional name in your 'majordomo_dont_reply' regex.

-- 
 Dave Wolfe


References:
Indexed By Date Previous: Re: archive update commands?
From: Jerry Peek <jpeek@jpeek.com>
Next: Re: majordomo on FreeBSD
From: Dave Wolfe <dwolfe@risc.sps.mot.com>
Indexed By Thread Previous: shlock problem
From: "James B. Byrne" <byrnejb@harte-lyne.ca>
Next: subscription bombing problem
From: "Lazlo Nibble" <lazlo@swcp.com>

Google
 
Search Internet Search www.greatcircle.com