Proxmox on centralized storage

[rsousa]

Hi,

We have been playing with proxmox for quite a while now and are very fond of it.
Unfortunately, as all of our storage is centralized (only available as NFS or iSCSI) we have been unable to deploy it in production.

Does any one have an idea of an hack that would allow us to use proxmox in such an infrastructure?

I' looking for something along the lines of «look at file zbr.pm in pve-manager, there is the code to copy/rsync/something and that's where your bound to have problems».

We don't expect this to be a «supported» configuration or anything

TIA

 

[rsousa]

So,
I've been looking at qemu_server's source.
qmigrate seems to hold the only NFS dangerous code.
A small patch to the rsync and «rm -f» areas should improve its work on NFS, no?

 

[dietmar]

I am working on more flexible storage model, but this will take a few months. But your hack is likely to work (for you), because kvm does not depend on the filesystem anyways.

- Dietmar

 

[mangoo]

Hi,

We have been playing with proxmox for quite a while now and are very fond of it.
Unfortunately, as all of our storage is centralized (only available as NFS or iSCSI) we have been unable to deploy it in production.

Does any one have an idea of an hack that would allow us to use proxmox in such an infrastructure?

I' looking for something along the lines of «look at file zbr.pm in pve-manager, there is the code to copy/rsync/something and that's where your bound to have problems».

We don't expect this to be a «supported» configuration or anything

TIA

If you use iSCSI, just set a proper path to iSCSI disk in /etc/qemu/<VMID>.conf, i.e.:

virtio0: /dev/disk/by-path/ip-192.168.112.66:3260-iscsi-iqn.2008-12.net.....-lun-1

Instead of pointing it to a file.

 

[rsousa]

Well, yes iSCSI is an option. But I'm afraid that doing things that way would negate (some of) the benefits of having a nice centralized console for your machines.
The work flow would become something like:
- Create,but don't start, a VM
- go to the storage interface and allocate a new volume to be exported via iscsi
- identify the volume on the server and bind it to the VM
- resume installing the VM

 

[d3vnull]

If you use iSCSI, just set a proper path to iSCSI disk in /etc/qemu/<VMID>.conf, i.e.:

virtio0: /dev/disk/by-path/ip-192.168.112.66:3260-iscsi-iqn.2008-12.net.....-lun-1

Instead of pointing it to a file.

Vista doesn't support virtio devices (please correct me if I am wrong but it hasn't worked for me). Still, a shared storage solution would work best in my environment as well. Like someone said earlier, if you patch qmigrate and remove certain parts you can use NFS. The only problem with this is that in the "Cluster" section of the administrative interface, you'll see a "cannot divide by 0" error. Other than that it works great.

 

[mangoo]

Vista doesn't support virtio devices (please correct me if I am wrong but it hasn't worked for me).

Use:

scsi0: /dev/disk/by-path/ip-192.168.112.66:3260-iscsi-iqn.2008-12.net.....-lun-1

or:

ide0: /dev/disk/by-path/ip-192.168.112.66:3260-iscsi-iqn.2008-12.net.....-lun-1

then...