[PVE-User] Two proxmox clusters sharing same storage

Chris Murray chrismurray84 at gmail.com
Thu Oct 16 12:54:29 CEST 2014


I think you're right and I'm assuming that's the way the process works, but I suppose I'm asking:

 

Why does it not enumerate the disks associated with a VM (and only those disks, used and unused), then remove them? Is there a technical reason not to do it that way?

e.g. why does:

rm /mnt/pve/nfsA/images/113/vm-113-disk-1.raw

turn into

rm /mnt/pve/nfsA/images/113/vm-113-disk-1.raw

rm /mnt/pve/nfsB/images/113/vm-113-disk-1.raw

 

What if a VM had two disks on two separate mountpoints and you remove one of the disks? You'd never expect it to remove the other one.

 

It's probably a good idea not to have two clusters share the same mount on the grounds of confusion and overlapping ID numbers, but deleting files which one cluster technically shouldn't even know exist doesn't make much sense to me.

 

Chris

 

From: pve-user [mailto:pve-user-bounces at pve.proxmox.com] On Behalf Of Eneko Lacunza
Sent: 16 October 2014 11:44
To: pve-user at pve.proxmox.com
Subject: Re: [PVE-User] Two proxmox clusters sharing same storage

 

Hi Chris,

Likely explanation is that when you remove a VM, it looks for all disks (used and unused) on all storages for removal. It is confused because you use the very same export point for both clusters.

I suggest you use different export points for different clusters, even if on the same FS. If you what to share templates, just hardlink or bind-mount the directory. This will allow you to share the space without confusing Proxmox.

Cheers
Eneko

On 16/10/14 12:37, Chris Murray wrote:

	Hi,

	 

	Can anyone explain why the below happens please and if it's meant to?

	 

	Suppose I have two clusters.

	 

	Cluster #1 uses NFS mount A (and B and C)

	Cluster #2 uses NFS mount B (and C)

	 

	On cluster #1, I create a VM with ID 113 on NFS A. Install an OS, all fine.

	On cluster #2, I create a VM with ID 113 on NFS B. Install an OS, all fine.

	 

	Both are working at this point.

	 

	On cluster #1, power down and remove VM 113.

	 

	Cluster #2's VM 113 hangs. Look on NFS B... the virtual disk has disappeared.

	 

	Why would the deletion of the VM from cluster #1 affect the disk on NFS B, despite it not being used for that VM? I understand that there's a mount to B from cluster #1, but the only file it should have deleted was on A.

	 

	I ran into this while migrating from one cluster to another. I was very careful to ensure that I wouldn't overlap two VMs with one file while I was DD'ing behind the scenes, but still suffered missing files when some of the original VMs were deleted because some old VMs were sharing the same ID numbers as new VMs.

	 

	Regards,

	Chris

	
	
	
	

	_______________________________________________
	pve-user mailing list
	pve-user at pve.proxmox.com
	http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user






-- 
Zuzendari Teknikoa / Director T├ęcnico
Binovo IT Human Project, S.L.
Telf. 943575997
      943493611
Astigarraga bidea 2, planta 6 dcha., ofi. 3-2; 20180 Oiartzun (Gipuzkoa)
www.binovo.es
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pve.proxmox.com/pipermail/pve-user/attachments/20141016/ecdaf4d3/attachment-0015.html>


More information about the pve-user mailing list