Jeremy Huntwork wrote:
> On 5/9/11 11:36 PM, Bryan Kadzban wrote:
>> Right, but the former /etc/sysconfig/network* was also default
>> configuration for how the bootscripts run.  Well, the network script
>> anyway.  :-)
> 
> Just ifup/ifdown (which are now in /sbin) and network-services/* which 
> is now in /lib/{network-}services - maybe I'm missing something?

The network-services/* stuff was what I meant by configuration.
(Especially the ifconfig.<device-name> file or directory.)

This is more or less bikeshedding at this point though, I think.

>> So networking goes down before the per-client sshd processes, and the
>> kernel isn't smart enough to kill those connections at interface-down
>> time.  :-(
> 
> Well, that's actually a feature and not a bug. If your connection gets 
> interrupted temporarily by forces outside your control, you want the 
> session to be restored without interrupt.

I'd be surprised to find systems that did "ip link set dev eth0 down"
outside root's control.  :-)

(That's what I meant as the possible trigger for killing connections,
but obviously the kernel doesn't do it, so never mind.)

>> In that case, this should work.  Maybe with a comment explaining why
>> it's necessary though.
> 
> Yeah, I should have pasted the two comments that Fedora has in their 
> scripts, too:
> 
>          # if we are in halt or reboot runlevel kill all running sessions
>          # so the TCP connections are closed cleanly

Looks fine.

Attachment: signature.asc
Description: OpenPGP digital signature

-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to