SCIENTIFIC-LINUX-USERS Archives

November 2013

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"~Stack~" <[log in to unmask]>
Reply To:
~Stack~
Date:
Sat, 30 Nov 2013 01:03:16 -0600
Content-Type:
multipart/signed
Parts/Attachments:
text/plain (3077 bytes) , signature.asc (265 bytes)
Hello everyone,

I have run into a problem with 6.4 that I can reproduce in my
environment that has had me stumped for a few hours. I found nothing
that appeared to directly relate to my problem when searching the
list/net. Maybe I have been staring at it too long or maybe its because
it is well past midnight, but I am confused and maybe someone else can
help. :-D

The problem: Servers don't get a DHCP IP after booting up.

The background:
I have a small test environment and I am working on building something
from "the base up". I want as small of a start install as I can get. I
PXE booted 4 servers with the following package selection of tools I
know I am going to need:

%packages --nobase
@Core
acpid
dbus
openssh-clients
%end

In the post section I have a script that configures a repo for other
software, but the only thing that is actually installed is
puppet+dependencies. There are ~230 packages installed when it boots the
first time.

I have done nothing special with the network configuration. It is just
DHCP. They get a DHCP IP for the PXE boot, then again during install,
however, when they boot up there is no request made to the server that I
can see. If I manually log in on the box and run 'ifup eth0' it pulls a
DCHP IP within seconds and all is well...until the next boot up when it
requires a manual start of the network again.

I manually have to run `ifup eth0` on every box on a fresh boot!

In my experimentation, I installed from a CD using the GUI and selected
"minimal install". I don't remember the exact number of installed
packages, but I think it was closer to 250 (I just remember it being a
few more; not a ton more) and both of those installs didn't have this
problem. They both got DHCP IP's at boot time.

I took a `rpm -qa` from one of them and one by one I added the missing
packages to one of the PXE booted servers before rebooting and trying
another package. I either missed a package or goofed something up
because that didn't solve the problem with no DHCP on boot. (I also
accidentally destroyed the original list so I will have to do another
install to get the package list difference :-[ ).

I don't see any errors or warnings or anything at all in the log files.
The service is set to start at boot and is running:
$ chkconfig network --list
network        	0:off	1:off	2:on	3:on	4:on	5:on	6:off
$ service network status
Configured devices:
lo eth0
Currently active devices:
lo eth0

And here is where I am stumped. The only things I can think of are: 1) I
missed a package somewhere 2) There is a package that has a file
configured in the minimal install that isn't configured when
manually/PXE installed.

This late at night, it is probably #1. If no one knows what the problem
might be, I will take two of the nodes (one PXE booted and one minimal
install from CD) and methodically try and see what differences there
are. I know that there are a number of packages installed when I remove
the "--nobase" option but I have not yet checked to see if my problem
goes away with that configuration. I will try that tomorrow as well.

Any thoughts on what package/configuration I need?

Thanks!
~Stack~



ATOM RSS1 RSS2