Subject: Network daemons using local sockets
To: None <current-users@NetBSD.ORG>
From: Martin Husemann <>
List: current-users
Date: 05/07/1999 07:03:07
I have a daemon that's needed for my ISDN network connections.

I fire this one up in /etc/netstart.local - that's what I thought is the
most appropriate place to do it.

Now this daemon uses a local socket to let a control program talk to
it. I placed that socket in /var/run - again what I thought is the
most appropriate place.

It gets created just fine when netstart.local is executed (from netstart)
at line 111 in /etc/rc. Then in line 142 it is deleted:

(cd /var/run && { rm -rf -- *; install -c -m 664 -g utmp /dev/null utmp; })

The solution in my case is simple, I'll delay the isdnd start until rc.local,
as (for now) it is a local daemon ;-)

But the general problem is there, and the question is even more intersting
if /var is on NFS.