Re: [Libguestfs] [rjones@redhat.com: Re: Suggestions on building VM disks from scratch]
On Sat, Apr 14, 2012 at 12:20:00AM +0100, Dale Amon wrote:
I might add that virt-rescue now gets much further than I have ever seen before... it shows a line [ 245.032513] vda: vda1 vda2
before hanging forever. So it at least has gotten to the point of recognizing the partitions on the virtual disk I want to access.
This hang is also characteristic of the ADFS problem. I don't think anything is going to change until someone fixes this kernel bug. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones New in Fedora 11: Fedora Windows cross-compiler. Compile Windows programs, test, and build Windows installers. Over 70 libraries supprt'd http://fedoraproject.org/wiki/MinGW http://www.annexia.org/fedora_mingw
On Sat, Apr 14, 2012 at 07:36:24AM +0100, Richard W.M. Jones wrote:
On Sat, Apr 14, 2012 at 12:20:00AM +0100, Dale Amon wrote:
I might add that virt-rescue now gets much further than I have ever seen before... it shows a line [ 245.032513] vda: vda1 vda2
before hanging forever. So it at least has gotten to the point of recognizing the partitions on the virtual disk I want to access.
This hang is also characteristic of the ADFS problem.
I don't think anything is going to change until someone fixes this kernel bug.
Is there more than one facit to this bug? The reports I've found all seems to say it is with a blank disk. In this case the virtual disk is partitioned and has a valid ext4 and a swap partition on it and only lacks the boot MBR and grub set up.
On Sat, Apr 14, 2012 at 01:08:20PM +0100, Dale Amon wrote:
On Sat, Apr 14, 2012 at 07:36:24AM +0100, Richard W.M. Jones wrote:
On Sat, Apr 14, 2012 at 12:20:00AM +0100, Dale Amon wrote:
I might add that virt-rescue now gets much further than I have ever seen before... it shows a line [ 245.032513] vda: vda1 vda2
before hanging forever. So it at least has gotten to the point of recognizing the partitions on the virtual disk I want to access.
This hang is also characteristic of the ADFS problem.
I don't think anything is going to change until someone fixes this kernel bug.
Is there more than one facit to this bug? The reports I've found all seems to say it is with a blank disk. In this case the virtual disk is partitioned and has a valid ext4 and a swap partition on it and only lacks the boot MBR and grub set up.
Although I can easily reproduce this bug, I'm not 100% clear about when it happens. Certainly, it happens with a blank disk. Also, it happens with a disk containing a single partition that has a filesystem on it. It hangs, and then (240 seconds later) has the same stack trace, so it appears to be the same bug. Why exactly it should happen in this case is not understood. I've not tried it with 2 x partitions on a disk. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into Xen guests. http://et.redhat.com/~rjones/virt-p2v
On Sat, Apr 14, 2012 at 01:08:20PM +0100, Dale Amon wrote:
On Sat, Apr 14, 2012 at 07:36:24AM +0100, Richard W.M. Jones wrote:
On Sat, Apr 14, 2012 at 12:20:00AM +0100, Dale Amon wrote:
I might add that virt-rescue now gets much further than I have ever seen before... it shows a line [ 245.032513] vda: vda1 vda2
before hanging forever. So it at least has gotten to the point of recognizing the partitions on the virtual disk I want to access.
This hang is also characteristic of the ADFS problem.
I don't think anything is going to change until someone fixes this kernel bug.
Is there more than one facit to this bug? The reports I've found all seems to say it is with a blank disk. In this case the virtual disk is partitioned and has a valid ext4 and a swap partition on it and only lacks the boot MBR and grub set up.
Then I have just now confirmed that this problem goes deeper and may be different than what people have been thinking. I tried virt-rescue just now on a working VM that uses partitions vda1-6, ie 3 primary and two extended, where one extended partition is a swap disk, all others are ext3 or 4 and the MBR is valid and functioning. It locks up at exactly the same point. I'm going to just confirm that the VM in question still works since I had not run it in a couple weeks. I will get back shortly with that confirmation.
On Sat, Apr 14, 2012 at 05:35:17PM +0100, Dale Amon wrote:
It locks up at exactly the same point. I'm going to just confirm that the VM in question still works since I had not run it in a couple weeks. I will get back shortly with that confirmation.
Nope. Started the same VM up remotely from virt-manager with nary a hiccup.
On Sat, Apr 14, 2012 at 05:54:32PM +0100, Dale Amon wrote:
On Sat, Apr 14, 2012 at 05:35:17PM +0100, Dale Amon wrote:
It locks up at exactly the same point. I'm going to just confirm that the VM in question still works since I had not run it in a couple weeks. I will get back shortly with that confirmation.
Nope. Started the same VM up remotely from virt-manager with nary a hiccup.
Whether the VM starts or not has no relationship to the kernel bug. libguestfs (including virt-rescue) boots its own appliance using the host kernel, which on Ubuntu suffers from this bug. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming blog: http://rwmj.wordpress.com Fedora now supports 80 OCaml packages (the OPEN alternative to F#) http://cocan.org/getting_started_with_ocaml_on_red_hat_and_fedora
On Sat, Apr 14, 2012 at 06:03:37PM +0100, Richard W.M. Jones wrote:
Whether the VM starts or not has no relationship to the kernel bug. libguestfs (including virt-rescue) boots its own appliance using the host kernel, which on Ubuntu suffers from this bug.
I might take it up with LKML then, because as you noted earlier the Debian kernel has the problem also, and Debian tends to be a relatively standard build. I'm wondering if perhaps the solution to my problem might be to use virt-manager to 'build' a machine with knoppix and attach it to the target. Or else pregen grub1, grub2 and lilo VM's that can attach to a disk on which grub1, grub2 or lilo needs to be installed to the boot block. I fear it will require at least two VM's because you cannot repair a grub1 host from a grub2 and vice versa.
On Sat, Apr 14, 2012 at 06:12:18PM +0100, Dale Amon wrote:
On Sat, Apr 14, 2012 at 06:03:37PM +0100, Richard W.M. Jones wrote:
Whether the VM starts or not has no relationship to the kernel bug. libguestfs (including virt-rescue) boots its own appliance using the host kernel, which on Ubuntu suffers from this bug.
I might take it up with LKML then, because as you noted earlier the Debian kernel has the problem also, and Debian tends to be a relatively standard build.
This happens on a standard, upstream Linux kernel, and has nothing to do with Debian or Ubuntu. Please follow the description on the BZ: https://bugzilla.kernel.org/show_bug.cgi?id=42778 You should start by CC-ing yourself on the bug. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://et.redhat.com/~rjones/virt-df/
participants (2)
-
Dale Amon -
Richard W.M. Jones