Bug #2900

Problem with onevm (and sunstone)

Added by Marcello Lodi about 7 years ago. Updated about 7 years ago.

Status:ClosedStart date:05/09/2014
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:Drivers - VM
Target version:Release 4.6
Resolution:worksforme Pull request:
Affected Versions:OpenNebula 4.6

Description

OS: CentOS 6.5 (x86_64)
One: 4.6

While trying to shutdown and/or migrate virtual machine, I realized that I can't.
In particular, when I execute the shutdown command (as oneadmin):
onevm 143 shutdown

I got these errors:

Fri May 9 12:20:10 2014 [ReM][D]: Req:1472 UID:0 VirtualMachineAction result SUCCESS, 143
Fri May 9 12:20:11 2014 [VMM][D]: Message received: LOG I 143 Command execution fail: /var/tmp/one/vmm/xen4/shutdown 'mkdir' 'one-frontend' 143 one-frontend

Fri May 9 12:20:11 2014 [VMM][D]: Message received: LOG E 143 shutdown: Command "sudo /usr/sbin/xl shutdown mkdir" failed: mkdir is an invalid domain identifier (rc=-6)

Fri May 9 12:20:11 2014 [VMM][D]: Message received: LOG E 143 Could not shutdown mkdir

Fri May 9 12:20:11 2014 [VMM][D]: Message received: LOG I 143 ExitCode: 2

Fri May 9 12:20:11 2014 [VMM][D]: Message received: LOG I 143 Failed to execute virtualization driver operation: shutdown.

Fri May 9 12:20:11 2014 [VMM][D]: Message received: SHUTDOWN FAILURE 143 Could not shutdown mkdir

It's clear that instead of using the domain name (in this case 'one-143'), it is used the 'mkdir' domain which is clearly an error.

Of course every command that a fired up from sunstone, also fails.

While I try to migrate the virtual machine with the command:

onevm migrate 143 virtual-mail.tng.iac.es

I got these errors:

Fri May 9 12:25:17 2014 [VMM][D]: Message received: LOG I 143 Command execution fail: /var/tmp/one/vmm/xen4/save 'mkdir' '/var/lib/one//datastores/0/143/checkpoint' 'one-frontend' 143 one-frontend

Fri May 9 12:25:17 2014 [VMM][D]: Message received: LOG E 143 save: Command "sudo /usr/sbin/xl save mkdir /var/lib/one//datastores/0/143/checkpoint" failed: mkdir is an invalid domain identifier (rc=-6)

Fri May 9 12:25:17 2014 [VMM][D]: Message received: LOG E 143 Could not save mkdir to /var/lib/one//datastores/0/143/checkpoint

Fri May 9 12:25:17 2014 [VMM][D]: Message received: LOG I 143 ExitCode: 2

Fri May 9 12:25:17 2014 [VMM][D]: Message received: LOG I 143 Failed to execute virtualization driver operation: save.

Fri May 9 12:25:17 2014 [VMM][D]: Message received: SAVE FAILURE 143 Could not save mkdir to /var/lib/one//datastores/0/143/checkpoint

144.xml Magnifier (5.12 KB) Marcello Lodi, 05/11/2014 03:11 PM

History

#1 Updated by Ruben S. Montero about 7 years ago

This seems to be pretty strange, 'mkdir' should be 'one-143' as mentioned. Could you send the output of
onevm -x 143, we are looking for the DEPLOY_ID element

#2 Updated by Marcello Lodi about 7 years ago

I instantiate another vm (144).

I send the result of the command: onevm show -x 144

Actually DEPLOY_ID is 'mkdir'

Have I set up something wrong in my configuration?

#3 Updated by Marcello Lodi about 7 years ago

Sorry I left an uncommented line inside the file
/var/tmp/one/vmm/xen4/deploy
(echo "mkdir .....").

Now it seems to deploy well, I mean with the DEPLOY_ID correctly set.
But when I try to deploy to another worker node I have problems due
to incorrect file permissions.
It seems to be almost the same problem of bug 2789; in particular the context
filesystem is created with 644 permission on the susnstone/workernode.
But when it is deployed (I suppose as root) in the other worker node is when it fails:

xl create -c deplyment.0Parsing config from deployment.0
DEBUG libxl__blktap_devpath 37 aio:/var/lib/one//datastores/0/155/disk.1
libxl: error: libxl.c:1871:device_disk_add: failed to get blktap devpath for 0x1f58b70

libxl: error: libxl_create.c:951:domcreate_launch_dm: unable to add disk devices

This is because disk.1 has not rwxrwxrwx permissions (only rw-r--r--)
How can I change it?
I tried to put DEFAULT_UMASK = 111 in oned.conf but with no success.

What am I doing wrong?
Cheers, Marcello

#4 Updated by Ruben S. Montero about 7 years ago

  • Status changed from Pending to Closed
  • Resolution set to worksforme

Also available in: Atom PDF