VM, missing disk

[MRisberg]

@Darkbeldin said in VM, missing disk:

@MRisberg What is weird is that we don't see xvda on the VM, could you show us the disks inside the VM?

Thank you @Darkbeldin for chiming in.

Disks and drivers inside of the VM:

And the result of this command on your host

xe vm-disk-list uuid=UUID of your VM

The command you provided returned this:

Here we can see XVDA (named million_C because it's the C drive within the VM).

So .. the hosts knows it's there, obviously, as I can boot the machine. Is there something other I can do than restarting both the toolstack and XO that you can come up with?

[MRisberg]

I built a new XO machine, compiled from sources with build f1ab62524 (xo-server 5.102.3, xo-web 5.103.0 as per 2022-09-21 17:00). Made no difference in regards to being able to see the XVDA drive.

The host knows the drive is there, XO doesn't.
I've already tried this: reset toolstack + restart the host.

Is there a way of resetting the interface / API on the host that XO talks to? (Not sure what API that would be.)

[Danp]

@MRisberg Do you have more that one SR? Can you show us the Disks tab for the VM in question?

[MRisberg]

@Danp

Hi.

The first image in the first post is the Disks tab of the VM.

These are my current SRs:

[Darkbeldin]

@Danp For me the disk tab for the Vm is the first screenshot of the first post:

And yes apparently he as at least two SR
Stor 15k adn Stor 10k
Unfortunately from there and without taking a look directly i'm a bit confused on what can be done perhaps he should try backuping the VM and restoring it?

[Danp]

@MRisberg OIC. I got confused when looking at this earlier. My suspicion is that the is-a-snapshot parameter has been erroneously set on the VDI.

[Darkbeldin]

@Danp Perhaps, but it will require to look at the chain in the host to see if your right to see if there is an child to this VDI.

[MRisberg]

@Danp said in VM, missing disk:

@MRisberg OIC. I got confused when looking at this earlier. My suspicion is that the is-a-snapshot parameter has been erroneously set on the VDI.

Interesting. Do you know of a way to check what parameters have been set for a VDI on the host?

Running the xe snapshot-list command did not return XVDA or million_C as having a snapshot, although XO indicates this with the camera icon.

Edit: Also, how did you get "is-a-snapshot" show up highlighted as in your post? Looked nice

[Danp]

@MRisberg said in VM, missing disk:

Do you know of a way to check what parameters have been set for a VDI on the host?

xe vdi-param-list uuid=<Your UUID here>

Also, how did you get "is-a-snapshot" show up highlighted as in your post?

With the appropriate Markdown syntax (see Inline code with backticks).

[MRisberg]

@Danp

Interesting, and a bit strange that XVDA (million_C) is-a-snapshot ( RO): true when earlier I wasn't able to list it as a snapshot at all. Oh, maybe the command makes a difference between VM snapshot and a VDI snapshot. Still learning. It returned:

XVDA:

XVDB:

Obviously there are some differences regarding the snapshot info but also million_C is missing info regarding vhd-parent. Not that I understand yet what that means.

Thank you.

[Danp]

@MRisberg I will defer to others on advising you on how to fix your situation. If the VM's contents are important, then I would be sure to make multiple backups in case of a catastrophic event. You should be able to export the to an XVA so that it can be reimported if needed.

[MRisberg]

@Danp

I appreciate your effort to help.
Seems like this is an odd one. Maybe the backup / restore strategy is the most efficient way forward here, as long as the backup sees both drives.

[Darkbeldin]

@MRisberg You take no risk at trying it you can keep the VM and restore it with another name.

[MRisberg]

While troubleshooting something else, backups also struggling, me and my friend found out that the host doesn't seem to coalesce successfully on the SR Stor 15K:

Sep 22 22:27:16 ninja SMGC: [1630] In cleanup
Sep 22 22:27:16 ninja SMGC: [1630] SR 35ec ('Stor 15K') (44 VDIs in 5 VHD trees):
Sep 22 22:27:16 ninja SMGC: [1630]         *3fc6e297(40.000G/85.000K)
Sep 22 22:27:16 ninja SMGC: [1630]         *4cce0cbb(100.000G/88.195G)
Sep 22 22:27:16 ninja SMGC: [1630]         *81341eac(40.000G/85.000K)
Sep 22 22:27:16 ninja SMGC: [1630]         *0377b389(40.000G/85.000K)
Sep 22 22:27:16 ninja SMGC: [1630]         *b891bcb0(40.000G/85.000K)
Sep 22 22:27:16 ninja SMGC: [1630]
Sep 22 22:27:16 ninja SMGC: [1630] *~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
Sep 22 22:27:16 ninja SMGC: [1630]          ***********************
Sep 22 22:27:16 ninja SMGC: [1630]          *  E X C E P T I O N  *
Sep 22 22:27:16 ninja SMGC: [1630]          ***********************
Sep 22 22:27:16 ninja SMGC: [1630] gc: EXCEPTION <class 'util.SMException'>, Parent VDI 6b13ce6a-b809-4a50-81b2-508be9dc606a of d36a5cb0-564a-4abb-920e-a6367741                            574a not found
Sep 22 22:27:16 ninja SMGC: [1630]   File "/opt/xensource/sm/cleanup.py", line 3379, in gc
Sep 22 22:27:16 ninja SMGC: [1630]     _gc(None, srUuid, dryRun)
Sep 22 22:27:16 ninja SMGC: [1630]   File "/opt/xensource/sm/cleanup.py", line 3264, in _gc
Sep 22 22:27:16 ninja SMGC: [1630]     _gcLoop(sr, dryRun)
Sep 22 22:27:16 ninja SMGC: [1630]   File "/opt/xensource/sm/cleanup.py", line 3174, in _gcLoop
Sep 22 22:27:16 ninja SMGC: [1630]     sr.scanLocked()
Sep 22 22:27:16 ninja SMGC: [1630]   File "/opt/xensource/sm/cleanup.py", line 1606, in scanLocked
Sep 22 22:27:16 ninja SMGC: [1630]     self.scan(force)
Sep 22 22:27:16 ninja SMGC: [1630]   File "/opt/xensource/sm/cleanup.py", line 2357, in scan
Sep 22 22:27:16 ninja SMGC: [1630]     self._buildTree(force)
Sep 22 22:27:16 ninja SMGC: [1630]   File "/opt/xensource/sm/cleanup.py", line 2313, in _buildTree
Sep 22 22:27:16 ninja SMGC: [1630]     "found" % (vdi.parentUuid, vdi.uuid))
Sep 22 22:27:16 ninja SMGC: [1630]
Sep 22 22:27:16 ninja SMGC: [1630] *~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*
Sep 22 22:27:16 ninja SMGC: [1630] * * * * * SR 35ecc7ae-4e98-58c1-5ed3-2c22c649bd32: ERROR

Maybe my problem is related to the host rather than XO? Back to troubleshooting .. and any input is appreciated.

[Darkbeldin]

@MRisberg That's mostly the issue apparently you lost a VDI from a chain there.
So the fact that the disk is now seen as a snashop can come from there yes.
But solving coalesce issue is not easy, how much do you love this VM?

[MRisberg]

@Darkbeldin
I actually solved the coalesce issue yesterday. After removing a certain VDI (not related) and a coalesce lock file, SR Stor 15K managed to process the coalesce backlog. After that the backups started to process again.

Due to poor performance I choose to postpone exporting or backing up the million VM. When I do, I'll see if I can restore it with both XVDA and XVDB.

I'll copy the important contents from the million VM before I do anything. That way I can improvise and dare to move forward in anyway i need to - even if I'll have to reinstall the VM. But first I always try to make an attempt at resolving the problem rather than just throwing away and redo from start. That way I learn more about the platform.

[Anonabhar]

@MRisberg Just wondering.. Where is the lock file? I might try that in the future for myself..

[Darkbeldin]

@Anonabhar Lock file should be at the root of the SR mount directory.

[MRisberg]

@Anonabhar

For me it was here:
On my (only) host: /run/sr-mount/35ecc7ae-4e98-58c1-5ed3-2c22c649bd32
(where 35ecc etc is one of my SRs)

There was a file named coalesce_ followed by a VDI UUID.

I moved this file out of the directory, rather than deleting it .. moving slowly.

[MRisberg]

After export and import I can actually see a reference to both drives. Although they seem to be disconnected.

I'll see what I can do about it.