Re: DHCP

> Yes, but I've noticed that Debian's network scripts also have a
> reasonable time-out too and I don't know if that is built-in or not.

The centos machine in question doesn't have a client script that I can
see. The ubuntu machine I have here doesn't have anything set either
- other than a few require lines its all commented out.

Which timeout are you referring to?

There is timeout, select-timeout, retry, and initial-interval. None
of these actually tell dhclient to completely give up from what I can
see.

Though reading the manual some more, it seems the initial-interval
won't triple forever - it will hit the retry time (defaults to 5
minutes) and then restart with the base initial-interval again. But
if that was the case my missing machine would have tried again. In
fact all my machine would have retried after 5 minutes which didn't
happen. So my reading of retry's effect on initial-interval is wrong.
 Unless there really is a point where the client gives up completely.

I'll have to setup a test bed to see the actual behavoir.

I've always been a proponent of having all machines, even static ones,
use DHCP. However I've always wondered what happens after a very long
(longer than the lease) dhcpd outage. Now I'm really motivated to
find out :-)
Received on Sat Mar 22 18:10:33 2008