Xen users - Jun 2008 - vm create test ="losetup /dev/loop0 /data/xen/domains/test/disk.img failed"

Dear mailing-list

after several attempts to start a xen linux-vm on a Fedora 8 XEN-System
(Quadcore Xeon-System with 8GB RAM and VT/PAE enabled), i now really
need your help.

I sucessfully created and installed a new debian etch-vm with the well
known debootstrap-script. I got 10 unused loop-devices which should be
more than enough for at least one xen-machine.

When i try to start the new debian-etch-machine with "*xm create
/etc/xen/test -c*" i always get the following error message:
-------------------------------
Using config file "./test".
Error: Device 2049 (vbd) could not be connected. losetup /dev/loop0
/data/xen/domains/test/disk.img failed
-------------------------------

*xen-hotplug.log:*
-------------------------------
Nothing to flush.
/data/xen/domains/test/disk.img: Permission denied
xenstore-read: couldn''t read path backend/vbd/2/2049/node
-------------------------------

The entire directory is world-accessible and writable, so there should
be no permission-problems with accessing the disk-images (VBD). I am
using the root-account to start the xen-vm.

*The configuration-file for the "test"-vm looks like this:*
-------------------------------
# Configuration file for the Xen instance vm01.example.com, created on
# Mon Jun 2 17:06:01 2008.
# Kernel + memory size
kernel = ''/boot/vmlinuz-2.6.21.7-3.fc8xen''
ramdisk = ''/boot/initrd-2.6.21.7-3.fc8xen.img''
memory = ''128''
# Disk device(s).
root = ''/dev/sda1 ro''
disk = [ ''file:/data/xen/domains/test/disk.img,sda1,w'',
''file:/data/xen/domains/test/swap.img,sda2,w'' ]
# Hostname
name = ''test''
# Networking
dhcp = ''dhcp''
vif = [ '''' ]
# Behaviour
on_poweroff = ''destroy''
on_reboot = ''restart''
on_crash = ''restart''
---------------------------------

An attempt to mount the disk-images to an unused-loop-device with
losetup works without any problems.

----------------------------------
[root@dmxen xen]# losetup /dev/loop0 /data/xen/domains/test/disk.img
[root@dmxen xen]# losetup -a
/dev/loop0: [0902]:192479239 (/data/xen/domains/test/disk.img)
----------------------------------

System details:
_Kernel:_ Linux dmxen 2.6.21.7-3.fc8xen #1 SMP Thu Mar 20 14:57:53 EDT
2008 i686 i686 i386 GNU/Linux
Xen-Version: Xen 3.1.2


Any help or information would be very welcome.

Thanks in advance

Christian



_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Wednesday June 18 2008 03:40:18 am Christian Nitschmann
wrote:
> When i try to start the new debian-etch-machine with "*xm create
> /etc/xen/test -c*" i always get the following error message:
> -------------------------------
> Using config file "./test".
> Error: Device 2049 (vbd) could not be connected. losetup /dev/loop0
> /data/xen/domains/test/disk.img failed
Maybe selinux? If it is set to enforcing, your images have to be 
in /var/lib/xen/images.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

I''ll second the suggestion of SELinux; you either need to:

a) switch it off (not ideal since you''ll lose the security benefits)
b) make sure the image is in the right place to satisfy it
c) educate SELinux about your preferred domain image path.

I''d also note that using tap:aio: instead of file: has been the
preferred
means of serving a block device from a file for a while now and might be 
worth trying.

Cheers,
Mark
> On Wednesday June 18 2008 03:40:18 am Christian Nitschmann wrote:
> > When i try to start the new debian-etch-machine with "*xm create
> > /etc/xen/test -c*" i always get the following error message:
> > -------------------------------
> > Using config file "./test".
> > Error: Device 2049 (vbd) could not be connected. losetup /dev/loop0
> > /data/xen/domains/test/disk.img failed
>
> Maybe selinux? If it is set to enforcing, your images have to be
> in /var/lib/xen/images.
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users


-- 
Push Me Pull You - Distributed SCM tool (http://www.cl.cam.ac.uk/~maw48/pmpu/)

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Thank you for the fast answer.
It seems that SELinux was the culprit. After disabling it, there were no 
problems with starting a xen-domu with disk-images in my own directories.

Greetings from Germany

Chris


Mark Williamson schrieb:
> I''ll second the suggestion of SELinux; you either need to:
>
> a) switch it off (not ideal since you''ll lose the security
benefits)
> b) make sure the image is in the right place to satisfy it
> c) educate SELinux about your preferred domain image path.
>
> I''d also note that using tap:aio: instead of file: has been the
preferred
> means of serving a block device from a file for a while now and might be 
> worth trying.
>
> Cheers,
> Mark
>
>   
>> On Wednesday June 18 2008 03:40:18 am Christian Nitschmann wrote:
>>     
>>> When i try to start the new debian-etch-machine with "*xm
create
>>> /etc/xen/test -c*" i always get the following error message:
>>> -------------------------------
>>> Using config file "./test".
>>> Error: Device 2049 (vbd) could not be connected. losetup /dev/loop0
>>> /data/xen/domains/test/disk.img failed
>>>       
>> Maybe selinux? If it is set to enforcing, your images have to be
>> in /var/lib/xen/images.
>>
>> _______________________________________________
>> Xen-users mailing list
>> Xen-users@lists.xensource.com
>> http://lists.xensource.com/xen-users
>>     
>
>
>
>   
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users