Launch VM (Compute/XenServer): Error: Could not stat device /dev/sda - No such file or directory.

Asked by Juan Perez

Hi,

When trying to boot vm, Status shows as 'active', but vm never boots up in XenServer, but the vm's glance disk does show up in XenServer.

(when run: cloudservers boot test --flavor=1 --image=11)
(when run: cloudservers list)

Logs pasted @ http://paste.openstack.org/show/606/

I tried different images, none of them Maverick, but they both had the same error.

It looks like this may be related to Bug (https://bugs.launchpad.net/nova/+bug/705790), but the fix was merged in 602, and I'm running revision 631.

Thank you.

Question information

Language:
English Edit question
Status:
Answered
For:
OpenStack Compute (nova) Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Salvatore Orlando (salvatore-orlando) said :
#1

Hi Juan,

Looking at the log you pasted, it seems the problem might be the same as bug #705790.
For a better assessment, It would be very helpful to know a few more details:
1) are you running the compute node on a maverick distro?
2) is the compute node running on a VM? If yes, is this VM running on the same hypervisor on which you want to launch the instance?
3) which log level are you currently launching nova with?

Thanks,
Salvatore

Revision history for this message
Juan Perez (juanperez) said :
#2

Hi Salvatore,

1) I"m using Lucid 10.04 on compute node
2) Yes, compute node is running on a VM on the same hypervisor on which i'm trying to launch an instance.
3) Logging appears to be full

Update, I was able to get past the error by changing the "--xenapi_remap_vbd_dev" entry in the nova.conf file to false from true:

--xenapi_remap_vbd_dev=false

But, I'm now seeing a similar error saying that permission denied to "/dev/xvda":

(nova.compute.manager): TRACE: Command: parted --script /dev/xvda mklabel msdos
(nova.compute.manager): TRACE: Exit code: 1
(nova.compute.manager): TRACE: Stdout: 'Error: Error opening /dev/xvda: Permission denied\n'

Thank you.

Revision history for this message
Salvatore Orlando (salvatore-orlando) said :
#3

Thanks for the update Juan.

Changing the flag to false makes sense in order to obtain the rigth prefix for devices.
I think your last problem might be related to access rigths on your compute node.
It seems the user which is running nova services cannot open the device for writing (the failure occurs when partitioning the VDI mounted in /dev/xvda).

One way to fix this is to update udev rules.
This rule should be added to /etc/udev/rules.d/50-udev.rules (default udev rules file, actual file might differ on your machine)

KERNEL=="xvd[a-z]", GROUP="$youruser", MODE="0660"

Hope this helps,
Salvatore

Can you help with this problem?

Provide an answer of your own, or ask Juan Perez for more information if necessary.

To post a message you must log in.