[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-users] CentOS 4.5 on SLES10SP1 no disk

  • To: Scott Serr <serrs@xxxxxxxxxxxx>
  • From: Nico Kadel-Garcia <nkadel@xxxxxxxxx>
  • Date: Sat, 14 Jul 2007 04:44:02 +0100
  • Cc: Henning Sprang <henning_sprang@xxxxxx>, xen-users@xxxxxxxxxxxxxxxxxxx
  • Delivery-date: Fri, 13 Jul 2007 20:37:47 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:user-agent:mime-version:to:cc:subject:references:in-reply-to:content-type:content-transfer-encoding; b=qtjvw3L6X1NSulSDuaeAlMyhRHXSOqEDG0uuAw8hI+iC5CSx6sNK7+j22IeZGWDSRjijdAxWLhMHb10vsEJbYyQInG7++xA7RIQYmP+gg+zQ13qEFJo3Etd/FJ7tTj3nWg4cRuPUN2q05DOXF50iLErnP2PBvpVLhk+nBh5YR9M=
  • List-id: Xen user discussion <xen-users.lists.xensource.com>

Scott Serr wrote:
Nico Kadel-Garcia wrote:
Henning Sprang wrote:
Scott Serr wrote:
I have a server running SLES10SP1 with Xen enabled.  I am able to fire
off SLED10 installs inside Xen VMs just fine.

I need to run (3) CentOS 4.5 VMs on it.

CentOS kernel is not creating block devices for hd or iso image. any ideas?

It doesn't seem to care if there is this line in my config file:

disk = [ 'file:/export/home/CentOS-4.5-x86_64-binDVD.iso,hdc,r' ]

Hmm, you're trying to boot an ISO in paravirtual mode?

AFAIK this isn't so easy as one would assume when having used other
virtualization technologies before.

If at all, you can _maybe_ expect it work halfways nice with HVM.

But you might rather try bootstrapping centos with yum, as described
(this is tested on Debian, it might or might not work on SuSE, I heard
various reports on success and failure on other distributions).

My experience with this approach is that yum hangs doing this on CentOS 4 and CentOS 5, on the "touch" commands run by various "%post" RPM function. You may need to set a lockfile with your build script, and set a little monitor widget to slap such touch processes in the head. Once you have a base minimum OS to your flavor, set it aside in a tarball for starting new configurations.

Or, if you're a complete weasel, you use the "mock" tool from Fedora Core to build full chroot cages, run "mock init", and do your upgrades inside the mock-built chroot.

Or, you might search the web - I used to know where I found this, but
have it not available now - there is a howto on how to start anaconda
inside a domU to make the installation, that should work with SuSE, too.

Take care to give enough(I don't know a number, a good guess is 256 if
on doubt) memory to the VM, otherwise Anaconda will likely hang in an
undefined state without any error message.
For CentOS 5, I find 512 works well. CentOS 4 was quite happy with 256: I haven't looked at the exact difference.

I'm giving the domU 512MB. Thank you both for your help. And I am just doing PV, forgot to say that.

What is weird though is when the kernel boots I never see anything about a hard drive either, not just the CD...

disk = [ 'file:/xen/dom1/disk0,hda,w' 'file:/xen/cd/centos4.iso,hdc,r' ]

I've created the disk0 with dd. I'd expect the kernel to mention that the disk has no partitions on hda, as I've seen before. I also tried the front end device name of xvda.
It only whines that way if you use something like disk = [ 'file:/xen/dom1/disk0,hda1,w' 'file:/xen/cd/centos4.iso,hdc,r' ]

You see the difference between hda, which describes a disk, and hda1, which describes a partition? They're also usually seen as sda or, with RHEL, xvda.

This page:
Has a full install from the network.

SuSE domU on SuSE dom0 is so easy...  I with I could get CentOS working.
Thanks for any additional help or hints.

Xen-users mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.