CentOS - Feb 2008 - Network Installation of CentOS disk image via PXE

Hello all,

I've deployed new servers - installing new CentOS servers via PXE
booting using its iso distribution stored on an NFS server.  For
certain server types;  I'd like to install custom applications into a
server and then generate an image of that server, and deploy again via
PXE to another group of servers.

Wanted to find out if anyone can forward any pointers to papers or links on:

1.  Best and (cheapest) way to create disk image (that can be used for
over the network installation over PXE) of an existing CentOS server
with all its custom apps and packages ... would this be dd?  would
this take a long time?

2.  Would like to find out if you can point me to a guide or doc -
that specifically describes this process;  most PXE install notes out
there describe the PXE config setup and assumes an ISO image (to
create a new server), as opposed to deploying a "ghosted" image of an
existing server.

3.  In a deployment of a "ghosted" image - would the DHCP
automatically request for a new IP address upon completion of the
installation on the target machine (since when I ghost the source
machine, the network information will contain that source's machine IP
address, MAC, etc. etc.)

Thanks in advance ..
-- 
best,

Vince

"vincenzo romero" <new2xen at gmail.com> wrote:
> Hello all,
>
> I've deployed new servers - installing new CentOS servers via PXE
> booting using its iso distribution stored on an NFS server.  For
> certain server types;  I'd like to install custom applications into a
> server and then generate an image of that server, and deploy again via
> PXE to another group of servers.
>
> Wanted to find out if anyone can forward any pointers to papers or links
on:
>
> 1.  Best and (cheapest) way to create disk image (that can be used for
> over the network installation over PXE) of an existing CentOS server
> with all its custom apps and packages ... would this be dd?  would
> this take a long time?
>
> 2.  Would like to find out if you can point me to a guide or doc -
> that specifically describes this process;  most PXE install notes out
> there describe the PXE config setup and assumes an ISO image (to
> create a new server), as opposed to deploying a "ghosted" image
of an
> existing server.
>
> 3.  In a deployment of a "ghosted" image - would the DHCP
> automatically request for a new IP address upon completion of the
> installation on the target machine (since when I ghost the source
> machine, the network information will contain that source's machine IP
> address, MAC, etc. etc.)
Unfortunately, I can't give you full details on this.  We did something 
very similar to what you're asking about in #1 and #2 where I used to 
work.  I used the setup for installing on systems but someone else did 
the initial setup.  You need to look at two things: PXE and custom 
kickstart.  I think Red Hat has some guidance on how to set up both.  If 
not, Google.

The idea is that the system that needs to get an installation is set to 
"boot from LAN" in the BIOS and the PXE server is also setup as a DHCP
server on the same LAN segment.  The system being installed does a 
normal boot from LAN which contacts the DHCP/PXE server.  The DHCP/PXE 
server provides the boot image using tftp from the first install CD and 
a kickstart file.  The kickstart file controls the installation as to 
which apps get installed.  The setup we used let us pick from several 
different possible kickstart files for different configurations so 
there's room for customizing the process.

Hopefully, that's enough of a start in the right direction that you can 
Google for the details.

As to #3, the idea is to have the source of the ghost image setup for 
DHCP booting and not be bound to a particular network card (HWADDR in 
the ifcfg-ethX file).  The target system gets the ghost image and then 
needs to reboot.  When it boots, it gets a new IP address via DHCP.

Cheers,
Dave

-- 
Politics, n. Strife of interests masquerading as a contest of principles.
-- Ambrose Bierce

vincenzo romero wrote:
> 
> Hello all,
> 
> I've deployed new servers - installing new CentOS servers via PXE
> booting using its iso distribution stored on an NFS server.  For
> certain server types;  I'd like to install custom applications into a
> server and then generate an image of that server, and deploy again via
> PXE to another group of servers.
> 
> Wanted to find out if anyone can forward any pointers to 
> papers or links on:
> 
> 1.  Best and (cheapest) way to create disk image (that can be used for
> over the network installation over PXE) of an existing CentOS server
> with all its custom apps and packages ... would this be dd?  would
> this take a long time?
> 
> 2.  Would like to find out if you can point me to a guide or doc -
> that specifically describes this process;  most PXE install notes out
> there describe the PXE config setup and assumes an ISO image (to
> create a new server), as opposed to deploying a "ghosted" image
of an
> existing server.
> 
> 3.  In a deployment of a "ghosted" image - would the DHCP
> automatically request for a new IP address upon completion of the
> installation on the target machine (since when I ghost the source
> machine, the network information will contain that source's machine IP
> address, MAC, etc. etc.)
> 
> Thanks in advance ..
We have PXE install in our environment that uses MS RIS to deploy
multiple RH distributions via kickstart using syslinux and the pxeboot
img included with the distros.

To recreate the setup completely on linux you will need:

1) DHCP server that supports PXE extensions

2) [Optional] PXE server for Linux to host multiple distros, if
you want to host just the one you could have DHCP point right to
the syslinux pxeboot loader.

3) TFTP server to host the initial boot images

4) WWW or FTP server to provide internal location to download the
distro RPMs (or you could use the Internet locations).

Here is a short How-To I found googling:

http://crashrecovery.org/CrashRecoveryKit/pxeboot/pxeboot.pdf

If you have a Win2k3 server license you could setup a Xen guest to
act as a RIS server too which would allow you to host Windows and
Linux distributions. You may have problems though with the DHCP/PXE
boot packets coming from the broadcast addresses to the guest, but
with tweaking I'm sure it could be made to work.

-Ross

______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.