Re: Problems with OpenBSD dhclient

From: Brooks Davis <brooks_at_one-eyed-alien.net>
Date: Thu, 14 Jul 2005 11:58:51 -0700
On Thu, Jul 14, 2005 at 07:27:10PM +0100, Robert Watson wrote:
> 
> On Thu, 14 Jul 2005, Kevin Oberman wrote:
> 
> >More serious is that I can't roam. When I move between APs, dhclient 
> >exits and I need to manually re-start it. I lose my SSH sessions. Ugh!
> >
> >Worse, I occasionally see my association drop momentarily when I am 
> >simply sitting and typing. Once again, dhclient dies and I must manually 
> >restart it and then re-establish my SSH and recover anything broken when 
> >the connection dropped. This is fairly serious! I don't understand what 
> >causes this, but it is infrequent which makes it hard to catch.
> >
> >It looks like killing dhclient when the interface drops is not a good 
> >idea. At very least, it needs to give a little time for re-association 
> >before dropping the DHCP client.
> 
> I'm experiencing similar or identical problems.  When at the British 
> Library the day before yesterday, slight blips in my association would 
> result in loss of my IP address, requiring manual restarting of dhclient.
> 
> My normal approach for wireless/etc is to settle on a link -- be it wired 
> ethernet or 802.11, and run dhclient on it to get things going.  When 
> done, I'll typically kill off dhclient.  This worked pretty well with the 
> old dhclient setup, but not so well with the new one.  Because I range 
> over a variety of network environments, from adhoc 802.11, wired ethernet, 
> to base station based 802.11, I tend to want to say "do this for now", and 
> have it stick until I move on.  Maybe what I need to change is software 
> configuration -- i.e., devd to re-launch dhclient.  But obviously what I'm 
> doing now isn't working, and results in a pretty poor user experience 
> (blips or roaming and suddenly the network is gone).

I'm seeing this as well.  I think we're going to need to handle wireless
and wired interfaces differently since their links work differently.

-- Brooks

-- 
Any statement of the form "X is the one, true Y" is FALSE.
PGP fingerprint 655D 519C 26A7 82E7 2529  9BF0 5D8E 8BE9 F238 1AD4

Received on Thu Jul 14 2005 - 16:58:56 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:38 UTC