Borja Sotomayor | 15 Dec 01:00 2005

Device 2049 (vbd) could not be connected. Backend device not found.

Hi everyone,

I'm running into a weird problem when running Xen 3.0. If I try to 
create new domains that use a single file-backed VBD (e.g. ttylinux-xen 
or a custom Debian image), everything works fine. However, if I try to 
create a domain that requires several VBDs, I get the following error 
message:

tp-c120:/vm/vm_images/vosg/GK # xm create -c osg-gk.config
Using config file "osg-gk.config".
Error: Device 2049 (vbd) could not be connected. Backend device not found.

[The config file osg-gk.config is included at the end of this e-mail]

Note that I get the same error message if I try to create a new domain 
with a custom Debian image (which works with a single VBD) and simply 
add a swap partition. Also, once I get this error, I can't create more 
domains (single or multiple VBD).

Along with the config file, I'm attaching the output from syslog, 
xend.log, and xm list --long. I noticed that the domain _does_ get 
created, but is placed in paused state. However, if I try to unpause it, 
the root partition misteriously disappears and the DomU kernel crashes 
(I'm also attaching the output of xm list --long after unpausing the domain)

Finally, here's some information on the machine and it's software:

     * Machine: 1U IBM e1350 eServer cluster node
     * IBM e325 Opteron Model 248 (2.2 GHz/1MB L2), 2x512 MB
     * Operating system: Enterprise Suse Linux 9 (SLES9) Service Pack 2
(Continue reading)

Ian Pratt | 15 Dec 01:42 2005
Picon
Picon

RE: Poweroff on xen-3.0-testing


> I get and install the xen-3.0-testing on my futur xenbox 
> (alias Bi-P3 on abit vp6 motherboard) but I have two problems.
>  
> 1/ I don't succeed in poweroff the dom0. When the poweroff is 
> started by "shutdown -h now" command the kernel power down 
> all disks devices but It seems the kernel don't call the 
> "acpi_power_off() alias pm_poweroff()" function for power 
> down the xenbox.
> Is it normal ? Do you have a patch for xen-3.0-testing ?

This is an open issue, but currently fairly low priority.

> 2/ The "ethtool" doesn't access to the data parameter of my 
> RTL8139 lan interface in the dom0 to active the WakeOnLan 
> power-on hardware fonction. Do you have any answer to that 
> failure ? Is it come from Brigde or XEN hypervisor ?

You should point ethtool at peth0 rather than eth0.
Ian
Satoshi Uchida | 15 Dec 02:32 2005
Picon

RE: Device 2049 (vbd) could not be connected. Backenddevice not found.

Hi. Borja 

What version do you use?
CD images?  Source download?

In Xen 3.0 CD which is download in 25 Nov., if one domain is constructed
using two and more than images, these VBDs conflict at same loop device.
This problem is fixed at lock mechanism before two weeks.

If would you like to download new patches ? 

Satoshi UCHIDA

-----Original Message-----
From: Borja Sotomayor [mailto:borja <at> borjanet.com] 
Sent: Thursday, December 15, 2005 9:00 AM
To: xen-users <at> lists.xensource.com
Subject: [Xen-users] Device 2049 (vbd) could not be connected.
Backenddevice not found.

Hi everyone,

I'm running into a weird problem when running Xen 3.0. If I try to
create new domains that use a single file-backed VBD (e.g. ttylinux-xen
or a custom Debian image), everything works fine. However, if I try to
create a domain that requires several VBDs, I get the following error
message:

tp-c120:/vm/vm_images/vosg/GK # xm create -c osg-gk.config Using config
file "osg-gk.config".
(Continue reading)

Borja Sotomayor | 15 Dec 02:43 2005

Re: Device 2049 (vbd) could not be connected. Backenddevice not found.

Hi,

> What version do you use?
> CD images?  Source download?

I'm using the 3.0.0 source download.

> In Xen 3.0 CD which is download in 25 Nov., if one domain is constructed
> using two and more than images, these VBDs conflict at same loop device.
> This problem is fixed at lock mechanism before two weeks.
> 
> If would you like to download new patches ? 

My guess is that the problem is not in Xen itself, but in some
configuration issue (I'm also wondering if the udev and hotplug packages
included in SLES9SP2 are a bit dated). Take into account that I have no
problem creating a domain with multiple images using the same Xen 3.0.0
on a different machine (with Ubuntu Breezy Badger, which is relatively
recent).

Cheers!
--

-- 
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Borja Sotomayor, University of Chicago
 Ph.D. Student, Department of Computer Science
 Hinds 016, 5734 South Ellis Avenue, 60637 Chicago, IL
GT4 Tutorial: http://gdp.globus.org/gt4-tutorial/
BorjaNet:  http://www.borjanet.com/   borja <at> borjanet.com
·····························································
         "Dis maschine vill run und run!"
(Continue reading)

Satoshi Uchida | 15 Dec 03:56 2005
Picon

RE: Device 2049 (vbd) could not be connected. Backenddevice not found.

>
> My guess is that the problem is not in Xen itself, but in some
configuration issue
> (I'm also wondering if the udev and hotplug packages included in
SLES9SP2 are 
> a bit dated). Take into account that I have no problem creating a
domain with multiple
> images using the same Xen 3.0.0 on a different machine (with Ubuntu
Breezy Badger
> , which is relatively recent).
>

Hi.

Ok,  I miss my comprehension.

>Dec 14 15:45:00 tp-c120 logger: /etc/xen/scripts/block:
>                                    Writing backend/vbd/1/2050/node
/dev/loop1 to xenstore.
>Dec 14 15:45:00 tp-c120 logger: /etc/xen/scripts/block: losetup
>                                   /dev/loop1
/vm/vm_images/vosg/GK/osg-gk-base.img failed

But,  I see that missing to allocate image file into loop back device.
your log shaw VBD 2050 ( /vm/vm_images/vosg/GK/swap1.img ) is allocated
loop1 and,
VBD 2049 ( file:/vm/vm_images/vosg/GK/osg-gk-base.img ) is trying to
allocate loop1.

Thank you.
(Continue reading)

Brian Lavender | 15 Dec 04:46 2005

Re: PXE installation of domU

On Tue, 2005-12-13 at 12:22 -0800, Kevin Fox wrote:
> Is there a way to support the execution of a PXE installation on a newly
> created domU?

Do you mean to do an installation? I don't think you need the pxe boot.
You just need to boot the kernel and the ramdisk for the installer? Only
thing is you will have to use a Xen kernel, and then you will have to go
into the ramdisk and replace the modules with the modules for the Xen
kernel.

Next problem I can think of is when you go to partition the disk. I am
not sure what would happen there. I wonder if you export a partition if
the installer will format it? 

Then it's going to have to install the Xen kernel into the VM at the
finish of install. 

But it would be cool to have an installer that would install a VM.

brian
--

-- 
Brian Lavender <brian <at> xensource.com>
Brian Lavender | 15 Dec 04:57 2005

RE: Device 2049 (vbd) could not be connected. Backenddevice not found.

On Thu, 2005-12-15 at 11:56 +0900, Satoshi Uchida wrote:
> >
> > My guess is that the problem is not in Xen itself, but in some
> configuration issue
> > (I'm also wondering if the udev and hotplug packages included in
> SLES9SP2 are 
> > a bit dated). Take into account that I have no problem creating a
> domain with multiple
> > images using the same Xen 3.0.0 on a different machine (with Ubuntu
> Breezy Badger
> > , which is relatively recent).
> >
> 
> Hi.
> 
> Ok,  I miss my comprehension.
> 
> >Dec 14 15:45:00 tp-c120 logger: /etc/xen/scripts/block:
> >                                    Writing backend/vbd/1/2050/node
> /dev/loop1 to xenstore.
> >Dec 14 15:45:00 tp-c120 logger: /etc/xen/scripts/block: losetup
> >                                   /dev/loop1
> /vm/vm_images/vosg/GK/osg-gk-base.img failed
> 
> But,  I see that missing to allocate image file into loop back device.
> your log shaw VBD 2050 ( /vm/vm_images/vosg/GK/swap1.img ) is allocated
> loop1 and,
> VBD 2049 ( file:/vm/vm_images/vosg/GK/osg-gk-base.img ) is trying to
> allocate loop1.
> 
(Continue reading)

Kate Keahey | 15 Dec 05:14 2005

Xen + Grid computing

In case there is interest in using Xen with Grid computing I just wanted 
to forward information about work we are doing within the Globus 
project. The message below is the annoucement of our tech preview 
release which came out today.

-------- Original Message --------

The Workspace Project at ANL is happy to announce technology preview 1.1
release of the Workspace Service.

The Workspace Service allows an authorized Grid client to dynamically
deploy and manage workspaces on the Grid. A workspace is an abstraction
of an execution environment that captures its software configuration
aspects as well as the resource quota assigned to it. In this
implementation, workspaces are implemented as Xen virtual machines,
deployed and managed based on WSRF protocols as implemented in Globus
Toolkit 4.

This release has a technology preview status: both interfaces and
implementation are likely to change to some extent. Information about
the project, software downloads, documentation and instructions on how
to join the workspace support list can be found at:
http://workspace.globus.org

A list of changes since last release is available at:
http://workspace.globus.org/vm/TP1.1/index.html

We welcome comments, feedback, and bug reports, even though the support
we can offer at this time is limitited. Looking forward to hearing from you!

(Continue reading)

Steven Ellis | 15 Dec 05:25 2005
Picon

Re: Centos + Xen 3.0 + Root on md = bad

larsks.21230602 <at> bloglines.com wrote:
>> I had the same problem. Fixed it with compiling the RAID modules
>>     
>
>   
>> in (CONFIG_MD_*).
>>     
>
>
>
> Taneli,
>
>
>
> Thanks for the information.
>
>
>
> That gives me two other questions:
>
>
>
> (1) This is for the folks at Xensource responsible for the binary distribution
> -- if md support under Xen requires the md drivers to be compiled into the
> kernel, why does the binary distribution from Xensource have them built as
> modules?
>
>
>
> (2) I've tried adding mdadm to the initrd, and it complains that
(Continue reading)

Sanjay Upadhyay | 15 Dec 06:42 2005
Picon

Re: xen and vlans

Same problem here in debian sarge,  and xen V3. Stuck !

On 12/15/05, Manuel da Costa <datashark <at> gmail.com> wrote:
> Hi all.
>
>
> What is the canonical way of getting xen to fully work with 802.1q (vlans) ?
> After (re)searching i found many ways but so far none is working.
>
> platform is gentoo (x86/32) and netwerk card a tg3. xen is v3.
>
> my config is as follows ...
>
> eth0 is  simply up (with no ip) and i created several eth0.x interfaces
> corresponding to several vlans.
> eth0.10 is de 'default' vlan (i.e. default gw).
>
> what i want is a way of each DomU connect to the world thru it's vlan
> bridge.
>
> for instance ... for a domU i defined   bridge=xenbr0.10. and in
> /etc/xen/xend-config.sxp i defined
>
> (network-script my-network-scripts)
>
>
> contents of /etc/xen/scripts/my-network-scripts bellow
>
> #!/bin/sh
> set -x
(Continue reading)


Gmane