PDA

View Full Version : DumU P2V, no drive found



Flako000
31-Jul-2013, 00:05
Hello, I'm trying to virtualize with xen a SLES11 SP1 (physical) in a SLESSP2 Dom0.
When you start the domU, this is not the virtual disk.
Modify fstab and menu.lst but sige without starting.
The truth that I'm a little confused, can not see that I'm missing.

The error is:


mount: devpts already mounted or /dev/pts busy
mount: according to mtab, devpts is already mounted on /dev/pts
Boot logging started on /dev/ttyS0(/dev/console) at Tue Jul 30 18:34:10 2013
resume device not found (ignoring)
Waiting for device /dev/hda2 to appear: [ 2.199923]
..............................Could not find /dev/hda2.
Want me to fall back to /dev/disk/by-uuid/fb5c6b0b-3688-43cd-8b20-d8fa3a0ba993? (Y)


Any suggestions to look at?


domU.cfg


name="bcs-bkp"
memory=1024
maxmem=1024
vcpus=2
on_poweroff="destroy"
on_reboot="restart"
on_crash="destroy"
#boot = "ba"
builder="hvm"
device_model="/usr/lib/xen/bin/qemu-dm"
kernel="/usr/lib/xen/boot/hvmloader"
disk=[ 'file:/mnt/util/xen-images/bcs/disco1,hda,w'
]
vnc=1
vncunused=1
acpi=1
serial="pty

Files fstab and menu.lst configuration - https://docs.google.com/file/d/0B8rkUchOnWAKbWlGTzFrNzFjZ1E/edit?usp=sharing
Console error log - https://docs.google.com/file/d/0B8rkUchOnWAKb2N4TWE0d1NSMFU/edit?usp=sharing
Log debuger of Dumu - https://docs.google.com/file/d/0B8rkUchOnWAKNWpHQktINHRoNjg/edit?usp=sharing

hanks for reading

jmozdzen
31-Jul-2013, 13:25
Hi Flako000,

I see you have set up a HVM - if there's no specific reason to do so, I'd suggest to go for a PVM, which will profit from optimized support in Dom0 and DomU kernel.

From your console error log, I can see that the kernel is using the SCSI layer for the disks (as is expected for the "newer" kernels), so probably the disk isn't known as "hda" but "sda":


[ 1.409882] SCSI subsystem initialized
[ 1.430893] scsi0 : ata_piix
[ 1.433835] scsi1 : ata_piix
[ 1.436113] ata1: PATA max MWDMA2 cmd 0x1f0 ctl 0x3f6 bmdma 0xc100 irq 14
[ 1.440538] ata2: PATA max MWDMA2 cmd 0x170 ctl 0x376 bmdma 0xc108 irq 15
[ 1.604331] ata1.00: ATA-7: QEMU HARDDISK, 0.10.2, max UDMA/100
[ 1.611916] ata1.00: 41943040 sectors, multi 16: LBA48
[ 1.618895] ata1.00: configured for MWDMA2
[ 1.622940] scsi 0:0:0:0: Direct-Access ATA QEMU HARDDISK 0.10 PQ: 0 ANSI: 5
[ 1.644048] Uniform Multi-Platform E-IDE driver

Regards,
Jens

Magic31
31-Jul-2013, 14:43
...
I see you have set up a HVM - if there's no specific reason to do so, I'd suggest to go for a PVM, which will profit from optimized support in Dom0 and DomU kernel.
..

True, with the added note that the guest will need to have the kernel-xen packages installed. Those are probably not there yet. Having the guest boot up as full virtual server should give the option to add in the needed packages so the guest can be reconfigured as paravirtual.

The first hurdle is indeed the change in device naming (hda vs sda or xvda).

Cheers,
Willem

jmozdzen
31-Jul-2013, 16:36
Hi Willem,


True, with the added note that the guest will need to have the kernel-xen packages installed. Those are probably not there yet. Having the guest boot up as full virtual server should give the option to add in the needed packages so the guest can be reconfigured as paravirtual.

The first hurdle is indeed the change in device naming (hda vs sda or xvda).

Cheers,
Willem

just for the records: you can actually create a PVM config where you specify kernel & ramdisk of your choice, it is not required to have those read from the DomU disk :) That way you can bring up you PVM and straighten things out, i.e. after a failed kernel update within your DomU.

Of course this is not for the casual Xen user, nor should one forget to copy the required modules to /lib/modules/.... inside DomU (unless no extras, on top of the modules loaded by the ram disk, are required).

Regard,
Jens

Magic31
31-Jul-2013, 19:54
Hi Willem,



just for the records: you can actually create a PVM config where you specify kernel & ramdisk of your choice, it is not required to have those read from the DomU disk :) That way you can bring up you PVM and straighten things out, i.e. after a failed kernel update within your DomU.

Of course this is not for the casual Xen user, nor should one forget to copy the required modules to /lib/modules/.... inside DomU (unless no extras, on top of the modules loaded by the ram disk, are required).

Regard,
Jens

Hey Jens,

Not something I've needed to do before... You need to teach me that craft sometime. Sounds quite useful indeed! :)

-Willem

Flako000
01-Aug-2013, 14:44
jmozdzen:
The choice of HVM is because planned/assumed it is more "fast" do ..
I agree that the PV option is better (thought to Physical -> HVM -> PV)

Probe with sda, but indicates the same error ..
In the same VM install SLES11SSP1, where installed as hda so I think it should be hda.

Also try to create all possible devices (hdaX, hdbX, sdaX, sdbX) with mknod in the bash-like console and mount the disk. But this also fails with the error:

mount-t reiserfs / dev/hda2 /aa
mount: / dev/hda2 is not a valid block device '.

So something is wrong .. maybe by this associated with the error "FATAL: Module xen_vbd not found." displayed on console.

Looking in there SLES11SP2 module, this in / lib/modules/3.0.13-0.27-default/updates/blkfront/xen-vbd.ko (Package xen-kmp-default-4.1.2_14_3.0.13_0.27-0.5.5.rpm)
but in my domU SES11SP1 no *vbd* module (HVM although I'm not sure that is necessary)


What else I can watch or read?

jmozdzen
02-Aug-2013, 10:46
Hi flako000,

jmozdzen:
The choice of HVM is because planned/assumed it is more "fast" do ..
I agree that the PV option is better (thought to Physical -> HVM -> PV)

Probe with sda, but indicates the same error ..
In the same VM install SLES11SSP1, where installed as hda so I think it should be hda.

Also try to create all possible devices (hdaX, hdbX, sdaX, sdbX) with mknod in the bash-like console and mount the disk. But this also fails with the error:

mount-t reiserfs / dev/hda2 /aa
mount: / dev/hda2 is not a valid block device '.

So something is wrong .. maybe by this associated with the error "FATAL: Module xen_vbd not found." displayed on console.

Looking in there SLES11SP2 module, this in / lib/modules/3.0.13-0.27-default/updates/blkfront/xen-vbd.ko (Package xen-kmp-default-4.1.2_14_3.0.13_0.27-0.5.5.rpm)
but in my domU SES11SP1 no *vbd* module (HVM although I'm not sure that is necessary)


What else I can watch or read?

a pure shot in the dark: have you tried setting

boot = "C"
in case that the bootloader code has some undocumented dependency (AFAIK, "boot" defaults to first hard disk, cdrom, "floppy disk", but I haven't checked if that is meant as "cda" or if it really is checking the type of defined disks)?

> So something is wrong .. maybe by this associated with the error "FATAL: Module xen_vbd not found." displayed on console.

Have you tried setting "ioemu" on the disk device (in the DomU config file)? The message is hinting that the HVM is trying to use PV access to the block device - which isn't available due to the missing module.

> I agree that the PV option is better (thought to Physical -> HVM -> PV)

Do you still have the hardware server available? You then may install the xen guest kernel on that machine (in addition to the regular kernel), re-clone the disk, adjust fstab etc and start the VM as a PVM...

Regards,
Jens

Flako000
02-Aug-2013, 14:05
jmozdzen

It seems that what was missing was to rebuild the initrd with the command:
mkinitrd-vk / boot/vmlinux-2.6.32.12-0.7-default.gz-i / boot/initrd-2.6.32.12-0.7-default-sda-M / boot/System.map-2.6.32.12-0.7-default - d / dev/sda2
edit /boot/grub/menu.lst to use the new initrd and root=/dev/hda2. with this worked

though not clear to me why it was necessary to create it again, in the INITRD_MODULES was defined xen-vbd module (INITRD_MODULES = "xen-vbd piix ata_piix ata_generic processor fan ide_pci_generic xen_vbd reiserfs edd")
Either because you need to use the parameter '-d / dev/sda2' (with /dev/hda2 does not work), being that just started using Linux hda2.

the option ioemu the probe, but indicates the same lack of module error

All this talk of problems modules, driver attempts to use pv. It could be because the source physical machine (which turn into hvm)
have installed the kernel-xen? (not used as dom0 physical equipment)

jmozdzen
02-Aug-2013, 14:49
Hi Flako000,


jmozdzen

It seems that what was missing was to rebuild the initrd with the command:
mkinitrd-vk / boot/vmlinux-2.6.32.12-0.7-default.gz-i / boot/initrd-2.6.32.12-0.7-default-sda-M / boot/System.map-2.6.32.12-0.7-default - d / dev/sda2
edit /boot/grub/menu.lst to use the new initrd and root=/dev/hda2. with this worked

good to know you got it working!


the option ioemu the probe, but indicates the same lack of module error
That is strange, indeed. From Xen's documentation: "ioemu means the disk will use ioemu to virtualize the VMX disk. If not adding ioemu, it uses vbd like paravirtualized guests." That PVM driver you were missing should not have been used at all when specifying "ioemu".

Nevertheless, as your system seems to be up & running now, I guess it's time to move on...

Regards,
Jens