Great Circle Associates Majordomo-Workers
(July 1996)
 

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

Subject: Re: Zero byte user lists
From: Paul Close <pdc @ cabernet . virtualvin . com>
Date: Tue, 9 Jul 1996 10:37:30 -0700 (PDT)
To: brozen @ netvoyage . net
Cc: majordomo-workers @ GreatCircle . COM (Majordomo Workers List)
In-reply-to: <Pine.BSI.3.94.960708205635.8016B-100000@digital.netvoyage.net> from "Brock Rozen" at Jul 08, 1996 08:57:13 PM

> -----
> From: Chris Ho <hocy@icg.resnet.upenn.edu>
> 
> Anyway, I've noticed that on heavily loaded systems, sometimes an 
> unsubscribe will truncate a list down to 0 bytes. I've tracked this bug 
> down to an error in the unsubscribe routine. Right now, the fix is
> putting a check that the prints were successful 
> 
> print NEW  $_ || &abort("blablabla");
> 
> and also before the original list is deleted:
> 
> if ($match_count == 1 && -s NEW) {
> 
> Anyone know of a better way to fix this, preferably a way to have 
> sendmail requeue the remove request for a retry?
> 
> -----

This is a fix I submitted way back in 1.93 days.  In fact, I put
in changes that also verified that the subscriber list had only
one address removed as a further sanity check (that's what the
match_count check is).  If this isn't in 1.94, something is seriously
wrong.  Maybe the patch just didn't get merged into '94 properly??

I'm just starting a new job, and don't have my majordomo sources off
tape yet, so I can't check myself.  I do have the patch I submitted way
back when.  I've attached it.

Paul Close
pdc@netcontents.com

Follow-Ups:
References:
Indexed By Date Previous: Re: Auto Administrative Messages
From: "James B. Byrne" <byrnejb@harte-lyne.ca>
Next: Re: Auto Administrative Messages
From: Brock Rozen <brozen@netvoyage.net>
Indexed By Thread Previous: Zero byte user lists
From: Brock Rozen <brozen@netvoyage.net>
Next: Re: Zero byte user lists
From: Brock Rozen <brozen@netvoyage.net>

Google
 
Search Internet Search www.greatcircle.com