Bug 879222

Or we should allow for deletion from multiple domain in the backend

Daniel, can we enable deletion of multiple copies simultaneously in a simple way?
If not, how difficult would it be to prevent multiple selection? (are there other operations which can run on multiple copies?)

(In reply to comment #2)
> Daniel, can we enable deletion of multiple copies simultaneously in a simple
> way?
> If not, how difficult would it be to prevent multiple selection? (are there
> other operations which can run on multiple copies?)

For deleting template disk, the webadmin is using RunMultipleAction with VdcActionType.RemoveDisk. I.e. deletion of multiple copies should work ui-wise.
There is a limitiation of deletion the last copy of a disk - it's inapplicable.
(the canDoAction error implies a lock validiation issue in the backend command)

More accurate reproduction scenario:

Create a template with several disks.
Go to Templates->select this template -->Disks sub tab.
For each disk there, select disk --> Copy (to another storage domain)
Now we have 2 copies of each disk - one on domain A, another on domain B.
Go to Storage sub tab of the same template.
Select same disk's different copies in different storage domains ( multiple selection with ctrl button)
Press "Remove"
The error message is displayed:
"Error while executing action: Cannot remove Virtual Machine Disk. Related operation is currently in progress. Please try again later."

Result: one of the disk copies is deleted, the other copy remains.

The bug is happening because of the nature of the exclusive lock.
The exclusive lock is taken on disk id.
If there is more than one disk instance on different storage domains, they still have the same id.
Thus, removing more than one disk instance (of same disk) will fail as they will try to take exact same lock. 
The first one will acquire the lock successfully and remove the disk.
The second disk - the command trying to remove it will try to acquire the lock , however the lock was already acquired during removal of first disk (because it's on same disk id) , hence removal of second disk will fail.

Evidence from the log:
2013-02-04 16:48:49,747 INFO  [org.ovirt.engine.core.bll.RemoveDiskCommand:51] (pool-10-thread-49) Lock Acquired to object EngineLock [exclusiveLocks= key: 9b8c7bbe-6a0e-4507-b5d8-c62cb14b4dfe value: DISK
, sharedLocks= key: a81afcce-20f5-4d8b-b9f9-84a8c9a22356 value: TEMPLATE

2013-02-04 16:49:02,640 INFO  [org.ovirt.engine.core.bll.RemoveDiskCommand:51] (pool-10-thread-50) Failed to Acquire Lock to object EngineLock [exclusiveLocks= key: 9b8c7bbe-6a0e-4507-b5d8-c62cb14b4dfe value: DISK
, sharedLocks= key: a81afcce-20f5-4d8b-b9f9-84a8c9a22356 value: TEMPLATE

A solution to be considered - change the exclusive lock to contain both disk id and storage domain id. This way it will still protect a single disk instance, however it will allow operations on multiple disk instances on different storage domains.

This solution should be expanded on all commands that use this lock the same way - RemoveDiskCommand, MoveOrCopyDiskCommand, AttachDiskToVmCommand

Maor,  I seem to recall we discussed this.  Is there anything you're doing to enable this?
Thanks.

(In reply to Ayal Baron from comment #7)
> Maor,  I seem to recall we discussed this.  Is there anything you're doing
> to enable this?
> Thanks.

Maybe you are referring to https://bugzilla.redhat.com/960169, but this is not the same issue.
The fix for BZ960169 was to use disk status for each image, which will not help to solve this issue.
The fix for that should be in the lock scope as Alissa mentioned in comment 5

This bug had requires_doc_text flag, yet no documentation text was provided. Please add the documentation text and only then set this flag.

This is very old and has not customer/user requests.
I'm closing until someone requests it.