[one-users] Problem with shared_lvm driver

[Alan Jurcic]

Hi,

While testing the shared LVM driver
(http://wiki.opennebula.org/shared_lvm) I've stumbled upon a slight
issue while trying to delete a VM with persistent image.

All other combinations of shutdown or delete with non-persistent or
persistent images work except deleting. The problem seems to be in the
order the scripts are executing:

Thu Apr  4 09:37:28 2013 [DiM][I]: New VM state is DONE
Thu Apr  4 09:37:28 2013 [VMM][W]: Ignored: LOG I 1012 Driver command
for 1012 cancelled
Thu Apr  4 09:37:28 2013 [TM][W]: Ignored: LOG I 1012 LV vg1/lv-one-317
in use: not deactivating
Thu Apr  4 09:37:28 2013 [TM][W]: Ignored: LOG I 1012 ExitCode: 0
Thu Apr  4 09:37:28 2013 [TM][W]: Ignored: LOG I 1012 delete: Deleting
/var/lib/one/datastores/0/1012
Thu Apr  4 09:37:28 2013 [TM][W]: Ignored: LOG I 1012 ExitCode: 0
Thu Apr  4 09:37:28 2013 [TM][W]: Ignored: TRANSFER SUCCESS 1012 -
Thu Apr  4 09:37:29 2013 [VMM][W]: Ignored: LOG I 1012 ExitCode: 0
Thu Apr  4 09:37:29 2013 [VMM][W]: Ignored: LOG I 1012 Successfully
execute virtualization driver operation: cancel.
Thu Apr  4 09:37:29 2013 [VMM][W]: Ignored: LOG I 1012 ExitCode: 0
Thu Apr  4 09:37:29 2013 [VMM][W]: Ignored: LOG I 1012 Successfully
execute network driver operation: clean.
Thu Apr  4 09:37:29 2013 [VMM][W]: Ignored: CANCEL SUCCESS 1012 -

Looks like nebula is cleaning up images before destroying the VM which
is kinda weird. It works with non-persistent images because you can
force removal of LV, but you can't deactivate it while it's being used.

Is this ordering different in more recent versions of OpenNebula (we're
using 3.4 atm)? Short of upgrading, are there any simple solutions to
this that I'm missing?

Regards,
Alan