Great Circle Associates Majordomo-Workers
(September 1996)
 

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

Subject: Re: Patch for 1.94 to avoid starvation when looping for lock
From: Jason L Tibbitts III <tibbs @ hpc . uh . edu>
Date: 27 Sep 1996 03:24:02 -0500
To: Dave Barr <barr @ math . psu . edu>
Cc: majordomo-workers @ GreatCircle . COM
In-reply-to: Dave Barr's message of Thu, 26 Sep 1996 10:16:20 -0400
References: <199609260135.SAA05454@toad.com> <199609261416.KAA25679@augusta.math.psu.edu>

>>>>> "DB" == Dave Barr <barr@math.psu.edu> writes:

DB> I agree a simple random 1-10 sleep is not sufficient.  Even on a few
DB> rapid bogus unsubscribes I get warnings from Majordomo on not finding
DB> L.Log -- apparently a race condition?  (or mabye shlock is warning on a
DB> no-problem condition?)

I recently sent a message ("Majordomo torture tactics" or somesuch)
detailing what happened when I ran the following:

#!/usr/local/bin/perl5 -w

foreach $i (1..100) {
  system("echo \"subscribe test-list testaddr$i\@host.dom\" | mail majordomo");
}
sleep 60;
foreach $i (4..100) {
  system("echo \"unsubscribe test-list testaddr$i\@host.dom\" | mail majordomo");
}

The gist was that I always get a few warnings for the first couple of
subscribes, until the random locking spaces the retries out a bit.  I don't
understand the locking code enough to figure out just why it's happening.
the exact warning is:

shlock: open("/usr/local/lists/lists/L.test-list"): No such file or directory

 - J<



References:
Indexed By Date Previous: Re: Locking and NT
From: Hugh Messenger <hugh@garply.com>
Next: Re: Looping, cascading Welcome and HELP messages (fwd)
From: Chan Wilson <cwilson@slurp.neu.sgi.com>
Indexed By Thread Previous: Re: Patch for 1.94 to avoid starvation when looping for lock
From: Dave Barr <barr@math.psu.edu>
Next: Strange alias problem?
From: Brian Abernathy <cba@wscoe1.atl.hp.com>

Google
 
Search Internet Search www.greatcircle.com