I've got a VM Server which used to have an unformatted physical disk (actually a RAID10 set on a RAID controller) which was allocated to a guest on it through VM Manager (although we'd not done anything with it yet, like mount it and put a filesystem on it).
Recently, I shut down the guest, edited it in VM Manager to remove the reference to the physical disk, and then shut down the VM Server upon which the guest (and the physical disk) was hosted. I then reconfigured the RAID volume to be a RAID5 set in the RAID card BIOS.
I then fired up the VM Server, logged into VM Manager and edited the guest to have a physical disk which was the new RAID5 volume. I fired up the guest, mounted the disk, and created a filesystem. So far so good.
Unfortunately, I noticed during allocation to the guest of the RAID5 volume as a physical disk that the RAID10 volume was also still allocatable, even though it didn't exist any more. I went to the "Servers and VMs" tab, clicked on the VM Server, changed to drop-down to "Physical Disks", and could see that the RAID10 volume was still visible. Clicking on the only active button, "Delete Filesystem", gives back "Selected Object not found". The "Edit" button is also active, but isn't any use.
Looking at the entry I see:
Thin Provision: No
Reserved by Server(s): N/A
Page83 ID: 36848f690ea14ff0018b7544208a919cc
Looking at the filesystem on the VM Server there is no "/dev/mapper/36848f690ea14ff0018b7544208a919cc".
So my question is: How do I delete the reference to this now non-existent physical disk in VM Manager, please?
You're right, I didn't. Apologies.
It's VM Manager 188.8.131.526. The fact that you mention that 3.2.2 fixes this bug amuses me greatly as I've just recovered from having to reinstall VM Manager after my 3.1.1->3.2.1 VM Manager update failed spectacularly. Attempting another update from 3.2.1->3.2.2 fills me with dread. Even though I dumped the OracleDB database for the OVS repository and went for the new, local, MySQL DB instead I'm worried I'm going to have to go through a clean install again.
But a fix for this bug, you say? In 3.2.2, you say? Nothing I can do in 3.2.1 about it?
Also, I note from the readme for the 3.2.2 update that the fix is:
16287721 Fixes the "Object Not Found" issue when a virtual disk file has been manually removed (or renamed)
This is a physical disk...
Edited by: Ben A on 08-Mar-2013 05:43
I agree with you 100 percent. I don't why on the world they keep revising distributions and requiring manual upgrades to get certain "bug" fixes. I hope they change this soon. Just seems so silly to me. I've been stable on 3.1.1 for months. I hate to even think of actually upgrading given this. I was lucky in that I stated testing with 3.0.1 and 3.1.1 came out during my testing. I would have hated to try to upgrade from 3.0.1 to 3.1.1 in a production environment.
I just ran a clean upgrade of the VM manager from 3.2.1 to 3.2.2 build 520 using MySQL. It went smooth. I do believe the "readme" details a bug fix associated with upgrading from 3.1.1 to 3.2.1.
Do a rescan physical disks from your VM server itself. Some people have had problems with servers rebooting doing this when there is I/O taking place on the server. Might want live migrate your guests before doing this. Rebooting should have cleared it once the Server synced with the VM Manager but the rescan from the VM Manager might get the server and the manager in sync.
Good to know the upgrade from 3.2.1 to 3.2.2 went smoothly. That gives me some confidence. I'll try it on Monday. 3.1.1 to 3.2.1 appeared to work, but it futzed something badly. I hope I never have to go through that again.
How do I do a rescan of the physical disks from the VM Server itself? I didn't think there was any way to do that except through the VM Manager GUI. As for live migration, I can't do that as there's no clustering and no shared filesystems/disks/etc. Everything is local. However, the guests aren't live yet, so I can shut them down whenever I like. Rebooting the VM Server doesn't seem to have cleared things, neither has rescans initiated by the VM Manager. So... How do I do this rescan on the VM Server?
Thanks for that. At the end all it said was:
0 new device(s) found.
0 device(s) removed.
So I doubt that was going to be the solution. The functionally seems completely broken as the new (replacement, existing) RAID volume isn't deletable either. Any other thoughts, please?
Very much appreciating the help.
I have now done an apparently flawless (a first for Oracle VM software) upgrade of Oracle VM Manager (and the Servers, too) from 3.2.1 to 3.2.2-520.
Unfortunately I still cannot delete old physical disk references from VM Servers via the VM Manager GUI. They're marked as "Warning" (in red), and also 'StorageDeviceOfflineEvent' "Physical disk is offline", but the option to delete the thing itself is greyed out.
Does anyone have any suggestions, please?