[ovirt-users] missing disk after storage domain expansion

Fri May 15 10:30:07 UTC 2015

Thank you for your reply!
I'm unsure if the disk contained any snapshots. I do not think so.
Is the register action a safe one to do in production system? I wouldn't
mess any of my existing running servers.

Here the relevant logs from the date
./engine.log-20150501.gz:2015-04-29 16:10:40,868  : disk creation
("hertz_disk5" newImageId = 4ab070c0-fb16-452d-8521-4ff0b004aef3)
./engine.log-20150512.gz:2015-05-04 17:22:42,691  : last  occurrence of the
imageid

Please note from
./engine.log-20150512.gz  2015-05-11 12:07:16,719 : creation of two disk OVF
store (??) on the same storage domain 7a48fe46-2112-40a4-814f-24d74c760b2d
right after the volume expansion

In the instructions.txt  what I did to enlarge the volume (and lose the
disk) (the guides is informational only, I followed the steps on a different
server)

Here some cats to thank you in advance
https://www.youtube.com/watch?v=nfTY41-zC7Y
AG

-----Original Message-----
From: Maor Lipchuk [mailto:mlipchuk at redhat.com]
Sent: Wednesday, May 13, 2015 11:02 PM
To: Andrea Ghelardi
Cc: users at ovirt.org
Subject: Re: [ovirt-users] missing disk after storage domain expansion

Hi Andrea,

First of all, the issue sounds quite severe, can u please attach the engine
logs so we can try to figure out how that happened.
second, does this disk contained any snapshots?
 If not, can you try to register it back (see
http://www.ovirt.org/Features/ImportStorageDomain#Register_an_unregistered_disk)

Regards,
Maor

----- Original Message -----
> From: "Andrea Ghelardi" <a.ghelardi at iontrading.com>
> To: users at ovirt.org
> Sent: Monday, May 11, 2015 12:01:17 AM
> Subject: Re: [ovirt-users] missing disk after storage domain expansion
>
>
>
> Ok, so,
>
> After _ a lot _ of unsuccessful approach, I finally connected to
> postegre DB directly.
>
> Browsing the tables I found “unregistered_ovf_of_entities” where there
> is a reference of the missing disk
>
>
>
>
>
> @ovf:diskId:4ab070c0-fb16-452d-8521-4ff0b004aef3
>
> @ovf:size:210
>
> @ovf:actual_size:210
>
> @ovf:vm_snapshot_id:2e24b255-bb84-4284-8785-e2a042045882
>
> @ovf:fileRef:16736ce0-a9df-410f-9f29-3a28364cdd41/4ab070c0-fb16-452d-8
> 521-4ff0b004aef3
>
> @ovf:format: http://www.vmware.com/specifications/vmdk.html#sparse
>
> @ovf:volume-format:RAW
>
> @ovf:volume-type:Preallocated
>
> @ovf:disk-interface:VirtIO
>
> @ovf:boot:false
>
> @ovf:disk-alias:hertz_disk5
>
> @ovf:disk-description:disk for SYBASE installation, on SAN shared
> storage
>
> @ovf:wipe-after-delete:false
>
>
>
> However, I’ve been unable to find any other helpful details.
>
> I guess the disk is not recoverable at this point?
>
> Any guru who has a good ovirt DB kwnoledge willing to give me some advice?
>
>
>
> Thanks as usual
>
> AG
>
>
>
>
> From: Andrea Ghelardi [mailto: a.ghelardi at iontrading.com ]
> Sent: Thursday, May 07, 2015 6:08 PM
> To: ' users at ovirt.org '
> Subject: missing disk after storage domain expansion
>
>
>
>
> Hi gentlemen,
>
>
>
> I recently found an error on 1 of my storages: it was complaining
> about no free space but the VM was running and disk operational.
>
> Since I needed to perform some maintenance on the VM, I shut it down
> and at restart VM couldn’t boot up properly.
>
> Checked VM via console and a disk was missing. Edited fstab (luckily
> this disk was not root but heck! It had a Sybase DB on it!) and
> restarted VM this time ok.
>
>
>
> Since the disk resides on the dstore with no space, I expanded the
> iSCSI LUN, then refreshed multipath on hosts, then resized PVs and now
> ovirt is showing the correct size (logs do not complain anymore on no free
> space).
>
>
>
> BUUUT
>
>
>
> Now disk is missing. It is not shown anymore on Disks tab nor anywhere
> else.
>
> Problem is that storage shows 214GB occupancy (size of the missing
> disk) so data is there but cannot find it anymore.
>
>
>
> Logs show original disk creation, errors from the lack of space,
> refresh of the storage size and then.... no more references on the disk.
>
>
>
> What can I do to find those missing ~210GBs?
>
>
>
> Cheers
>
> Andrea Ghelardi
>
>
>
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
-------------- next part --------------
We wanted to expand an iSCSI volume on the Compellent SAN and in turn expand the datastore in Ovirt. This feature has gone from being unsupported to supported (with manual intervention) in Ovirt 3.4 and above. It will at somepoint become a fully automated process.

1) Expand the volume on the Compellent. In the GUI right click the volume --> Expand

2) In Ovirt Admin Portal put the storage domain into maintenance mode.
Ovirt Admin Portal --> Data Centers --> Select the DC --> In the lower panel select storage --> Maintenance
When complete you should see a spanner icon next to the storage domain.

3) In the Ovirt Admin Portal --> Hosts tab Identify the node that is the SPM manager and start the next steps on that host first.

4) Verify the current disk size in multipath:
 multipath -ll | grep -A 11 36000d310004edc00000000000000002d
In this case the multipath volume shows the old size (21G). We have expanded the volume to 25G.

5) Rescan the iscsi interfaces to resize the underlying LUN:
[root at primitivo ~]# iscsiadm -m node -R -I iface0
[root at primitivo ~]# iscsiadm -m node -R -I iface1

6) Now we can check that the underlying LUN's have been expanded.
Take the devices from the multipath volume and check the size:
[root at primitivo ~]# fdisk -l | grep /dev/sdbf
We can see the underlying LUN shows the expanded size of 26G. Repeat this for the other devices (sdbi, sdbg, etc).

7) Resize the multipath volume:
[root at primitivo ~]# multipathd -k"resize map 36000d310004edc00000000000000002d"
You should now be able to verify the resize with multipath -ll:
[root at primitivo ~]# multipath -ll
Volume now shows the new size of 25G.

8) Resize the LVM volume. The physical volume name should match that from the mulitpath name:
Start with the physical volume:
[root at primitivo ~]# pvresize --setphysicalvolumesize 25G /dev/mapper/36000d310004edc00000000000000002d
This should come back saying "1 physical volume resized".
Now scan the volume groups:
[root at primitivo ~]# vgs

9) Now repeat steps 4,5,6 and 7 for the remaining nodes in the cluster. Once all nodes have been completed take the volume out of maintenance:
Ovirt Admin Portal --> Data Centers --> Select the DC --> In the lower panel select storage --> Activate
Once the domain comes back online it should show the new size. Guest disks can then be resized if required.

BUUUUT

very last step, no more disk to resize... :/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: engine.log-20150512.gz
Type: application/octet-stream
Size: 1227052 bytes
Desc: not available
URL: <http://lists.ovirt.org/pipermail/users/attachments/20150515/f4884934/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: engine.log-20150501.gz
Type: application/octet-stream
Size: 593320 bytes
Desc: not available
URL: <http://lists.ovirt.org/pipermail/users/attachments/20150515/f4884934/attachment-0003.obj>