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

Re: [Xen-users] How to change VM from para-virtualization to full-virtualization?


  • To: xen-users@xxxxxxxxxxxxx
  • From: Alexandre Kouznetsov <alk@xxxxxxxxxx>
  • Date: Fri, 09 Nov 2012 10:54:24 -0600
  • Delivery-date: Fri, 09 Nov 2012 16:55:31 +0000
  • List-id: Xen user discussion <xen-users.lists.xen.org>

Hello, Jing.

(Sorry if this is repeated. I don't see this mail I sent since yesterday in the list, so I re-send it.)

It's something wrong with your e-mail formatting, the screenshots you attached are not displayed inline.

Please consider not using HTML formatting when sending mails to the list. If you wish to share the output you get from your system, post or attach it as plain text.

El 08/11/12 01:53, Jerry Gao escribió:
*From: *Jing Chen
*Sent: *Thursday, November 08, 2012 2:14 PM
      This is Jing and I work with Jerry for this issue. I saw your
email and do some test following your instruction. As our case, I met
lots of issues since there are lots of track as you said. I list them
here and need your help to check.
Seems it is a little bit longger. But your help is grateful.

I create a PV for DomU and install the SUSE linux 11 SP2. my
configurations are as below.
1) partition.
I guess this is the screenshot showing "expert partitioner". As I understand it, this is how your PV guest looks initially and you use /dev/xvda2 as LVM physical volume. /dev/xvda2 seems to be a true partition within /dev/xvda, to you already have a partitioned block device for you HVM.

2) /etc/fstab file
Looks fine to me. Since your guest uses LVM, the device names are expected to remain unchanged after you switch to HVM, except /dev/xvda1 for /boot.

3) /boot/grub/menu.lst
     in PV, grub has been installed. The option 0 is the one work
normally in PV. I add the second option for HVM.
Note, that since your /boot is on separated partition, the path to kernel and initrd are wrong. As you specify it, they would be searched under /boot/boot. Remove the "/boot" prefix from pathnames of kernel and initrd images, your files are on "root" of your /boot partition.

4) the conf file in Dom0 is as below.
As I see, your DomU's xvda resides on your Dom0's /dev/sda4. This is not something I would recommend becouse of risk of confusion, one between many others). But should work if you are careful enough.

As a note, personally I use LVM within Dom0 (not DomU) and export multiple logical volumes as the DomU needs them. This instead of exporting a single block device and manage it's volumes within DomU.

To chagne PV to HVM. I do the following modification.
5) change conf file in Dom0
boot="d" should make you boot from CDROM, I'm not sure if that's your intention. In any case, it's good idea to boot a liveCD first, specially to take a look how the storage is presented to the DomU and make sure the boot loader is correctly installed.

So, does it boots from ISO?

6) change /etc/fstab as below
Looks good, but will work only if your LVM infrastructure is correctly recognized.

7) change /etc/mtab as below
/etc/mtab is usually auto-generated, in rare cases it needs to be modified by hand. In some occasion I had to do it to trick grub installer to place it's code in the correct place.

Then. I start up the HVM using xm create command. And the error is as below.
   I try to change root parameter in menu.lst, such as "root=/dev/hda2".
But it still can not startup saying can not mount /dev/hda2 to root.
Actually, it complains about mounting /dev/vg69/root. Not /dev/hda2, which is your physical volume, not root filesystem. (Am I looking to a wrong place?)

You have a rather complex storage setup. One of the benefits of virtualization is that you can hide this complexity from the VM, keeping it very simple, very small, very stable. Caesar's to Caesar, and infrastructure management to Dom0.

In this case, I bet initrd does not have LVM support, that's why it can't see /dev/vg69/root. Please consult your distribution documentation, about how to re-build initrd. Surely it's a very automated process.

Please give your advise what can I do.
Consider this steps:
1. Boot using the VM using your distribution's live CD in some "rescue" mode. 2. Make sure your /dev/hda, and all it's partitions and logical volumes are readable and mountable. 3. Threat this case as a normal, not virtualized machine that got something wrong with it's booting. Suse documentation shall cover it.


--
Alexandre Kouznetsov


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users


 


Rackspace

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