From owner-cvs-etc  Mon Jun  2 11:05:03 1997
Return-Path: <owner-cvs-etc>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id LAA02659
          for cvs-etc-outgoing; Mon, 2 Jun 1997 11:05:03 -0700 (PDT)
Received: from spinner.dialix.com.au (spinner.dialix.com.au [192.203.228.67])
          by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA02639;
          Mon, 2 Jun 1997 11:04:56 -0700 (PDT)
Received: from spinner.dialix.com.au (localhost.dialix.com.au [127.0.0.1])
          by spinner.dialix.com.au with ESMTP id CAA27545;
          Tue, 3 Jun 1997 02:03:38 +0800 (WST)
Message-Id: <199706021803.CAA27545@spinner.dialix.com.au>
X-Mailer: exmh version 2.0gamma 1/27/96
To: =?KOI8-R?B?4c7E0sXKIP7F0s7P1w==?= <ache@nagual.pp.ru>
cc: Adam David <adam@veda.is>, cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG,
        cvs-etc@FreeBSD.ORG
Subject: Re: cvs commit: src/etc rc 
In-reply-to: Your message of "Mon, 02 Jun 1997 21:48:38 +0400."
             <Pine.BSF.3.96.970602214748.22454A-100000@lsd.relcom.eu.net> 
Date: Tue, 03 Jun 1997 02:03:35 +0800
From: Peter Wemm <peter@spinner.dialix.com.au>
Sender: owner-cvs-etc@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

=?KOI8-R?B?4c7E0sXKIP7F0s7P1w==?= wrote:
> On Mon, 2 Jun 1997, Adam David wrote:
> 
> > PIDs gathered from /var/run should always be checked for validity before
> > killing anyway. The old process could have left a stale pidfile without a
> > new process being started. This can happen at any time, not only at reboot.
> 
> It is impossible.

Well, not quite "impossible", but it's not exactly simple for scripts to 
deal with..

Gated uses a flock() lock on it's pid file as the "lock", the existance of 
the file does not imply that it's valid.  So, things like gdc etc try and 
do a conflicting lock request which should succeed if gated is not runni
ng, and should fail if it is there.  But doing this for other daemons 
requires code changes in each one.

The problem with gated in particular was that if it was started with the 
network code, ie: before NFS mounts, it would put it's pid file in /var/
run.  However, after the NFS mounts took place, /var/run used to be 
cleared.. (in case /var was nfs mounted I guess).  So, gdc was not able to 
talk to the running gated.

Anyway, for innd, I always put the pid file in the newslib dir, right next 
to the active file since it was important to innd that it not "go away" on 
boot.

> -- 
> Andrey A. Chernov
> <ache@null.net>
> http://www.nagual.pp.ru/~ache/

Cheers,
-Peter