Why not use PRE_DOWN for this? You can write a script in the style of the firewall example (/usr/share/doc/netcfg/contrib/iptables.hook) and even enable it per interface (put PRE_DOWN in /etc/network.d/interfaces/[interface]) if you don't want to specify it per profile (your profiles may override it).

Excellent idea, jouke. I took a stab at it in the attached file but isn't very elegant and on my box, it doesn't work. I chmodded it to +x and placed in the location in its header. Upon issuing a `shutdown -r now` to the server, my terminal just hung... other times it hung for 10-20 sec and then reported " Write failed: Broken pipe." This is in contrast to when I execute the function manually. Then after about 3 sec, the connection is simply closed.

# /bin/ps aux | /usr/bin/grep ssh | /usr/bin/grep -v grep | /usr/bin/awk {'print $2'} | /usr/bin/xargs -r kill
Connection to mars closed by remote host.
Connection to mars closed.

I assume you have tested that line outside netcfg.
Does it work when placed literally in PRE_DOWN for a given profile?
PRE_DOWN='/bin/ps aux | /usr/bin/grep ssh | /usr/bin/grep -v grep | /usr/bin/awk {"print $2"} | /usr/bin/xargs -r kill'

Yes! This is excellent. Three questions:

1) The string of commands proposed isn't really elegant. Is there a more "correct" method that you're aware of to accomplish this? I spend a good 45 minutes googling sshd-native commands to force a logout of all users but came up empty.

2) Perhaps we can edit the included example profiles to include this or a similar line so users do not need to rely on the wiki and have these in the Arch package. How do you feel about this solution?

3) Is this really a netcfg issue? Perhaps that or a similar command should really go into the sshd.service or .socket files?

@jouke - Any thoughts on #1 and #2?

1) Try more googling, there are some alternatives. pkill seems to do it.

2) No, this is typical wiki material.

3) No, it should be part of ExecStop in the sshd service. In fact, if sshd behaves nicely, the spawned instances would belong to the right process group. All that would matter in that case, is that sshd is brought down before netcfg.

@jouke - Thanks for the pkill suggestion+ExecStop. Works wonderfully. Safe to close this task I think.

https://bugs.archlinux.org/task/32251