Re: DHCP failing in FC1

Manuel Carro wrote:

Rick,

Check /etc/sysconfig/network-scripts and verify that that the ifcfg-DHCP
(VERY bad name...see below) specifies "DEVICE=eth0:3". If not,
dhcpclient won't work since no DEVICE has been defined.

   Thaks for the advices.  Yes, the device is specified as you say.

The scripts in /etc/sysconfig/network-scripts should be named after
the interface it affects--NOT the protocol used to define its IP
address, e.g. "ifcfg-eth0:3", NOT "ifcfg-DHCP".

I'll try that; if it works I'll stick to it.

   However, while I understand that this setup might be needed
because the scripts currently require it, I do not see any deep reason
why it _must_ be so.  If the device is specified inside the file, then
the file may have any identifier you like.  In fact, I have
'ifcfg-Home' (a static IP for home) and 'ifcfg-Office' (ditto for the
workplace), and both work just right by issuing "ifup Office" and
"ifup Home" (which I usually do from the graphical interface).  And
this makes sense because it decouples the situation in which I am from
the connection method I use.  As soon as I get a wireless link at home
I'd change ifcfg-Home to start up the wireless interface, and
everything else (e.g., any startup script I may have or my partner's
routine to read email) should remain the same.  I think that this is
the right approach.

As I understand it, and from judicious reading of /etc/init.d/network, the system automatically reads the names of all the /etc/sysconfig/network-scripts/ifcfg-XXX files to find out which interfaces are configured and should be started.

I note that Christopher has posted a comment related to profiles for use with users and networking, but historically having one of the files above for each interface (and properly configured) will get that interface configured at boot time.

NOTE: If you use the network configuration tool ( redhat-config-network ( soon to be system-config-network)) it will create the file for you and save it. Then a boot will automatically start that interface unless you have it set the activate at boot option to no. I have not tried the profiles, but I assume that is where you can override the boot options and specify which interfaces are activated for each profile.

Again, if renaming the file to ifcgh-eth0:3 solves my problem,
that's fine. But if this is the only reason why it is not working
now, I still believe this is a bug in the scripts.

   OTOH, please developers do not take this as a criticism!  It is
very hard to set things right (I develop software myself) and you all
are doing a _great_ job.

In each file is a "DEVICE=" definition which should also specify the
device.

Then, there is redundancy. The name of the file must match
information that is also _inside_ the file.

       MCL
________________________________________________________________
[...] put Windows back into its place as an overpriced Nintendo.