JovianDSS recovery from a major storage failure - open-e/JovianDSS-Proxmox GitHub Wiki

In this guide we will show how to configure backup of Proxmox container and VM data located on JovianDSS storage.

For this purpose we will use 2 JovianDSS storage's:

Pool Pool-0 has 2 virtual IP's: 192.168.21.100, 192.168.31.100

1. Production node with pool Pool-0 with virtual IP's 192.168.21.100, 192.168.31.100
2. Backup node with pool Pool-0 with virtual IP's 192.168.22.100, 192.168.32.100

Make sure to introduce them to your proxmox as storage pools in /etc/pve/storage.cfg according to configuration guide

In demo setup it looks like this:

Pool present

Virtual IP's present

REST API is ON

/etc/pve/storage.cfg

joviandss: jdss-Production-node-Pool-0
        pool_name Pool-0
        user_name admin
        user_password admin
        content images,rootdir
        ssl_cert_verify 0
        control_addresses 192.168.21.100
        data_addresses 192.168.31.100
        path /mnt/pve/jdss-Pool-0
        shared 1

Pool present

Virtual IP's present

REST API is ON

/etc/pve/storage.cfg

joviandss: jdss-Backup-node-Pool-0
        pool_name Pool-0
        user_name admin
        user_password admin
        content images,rootdir
        ssl_cert_verify 0
        control_addresses 192.168.22.100
        data_addresses 192.168.32.100
        path /mnt/pve/jdss-Pool-0
        shared 1

To check that both storages configured properly, you can check proxmox server web page.

Herejdss-Production-node-Pool-0 and jdss-Backup-node-Pool-0 are online.

or use command line:

root@proxmox-demo:~# pvesm status
Name                               Type     Status           Total            Used       Available        %
jdss-Backup-node-Pool-0       joviandss     active       499122176        26214400       472907776    5.25%
jdss-Production-node-Pool-0   joviandss     active       266338304        46137344       220200960   17.32%
local                               dir     active        14145416         2831516        10573548   20.02%

Now when everything is working we would setup backup task for VM 100(windows-demo).

To do this we will create dedicated zvol on Backup node and set Production node to store back up information on it.

Go to Storage tab of Backup node and create backup-vm-100-disk-0 zvol. Name is arbitraty, but it would be better if this name will be easily associated with original volume.

1. Go to Storage/iSCSI Targets tab of Production node

2. Find zvol v_vm-100-disk-0 attached to respective target

3. Click setting for v_vm-100-disk-0

4. Select Add to backup task

5. Specify appropriate retension-interval plans

6. Set Destination server field to Backup node by providing its IP address in. In this case it is

7. Set Resource path as Pool-0/backup-vm-100-disk-0 or other zvol name if you named zvol backup-vm-100-disk-0 differently.

8. Go through other configurations

9. Check Production node Storage menu Snapshot tab. v_vm-100-disk-0 includes A and B icons.

• A for Auto-snapshots
• B for Backup functionality.

Imagine that something bad happens and virtual machine or its data gets corrupted, encrypted or entire server was damaged by fire or other disaster. In that case you can restore your virtual machine from back up that was made on Backup node. All you need to do is to clone backuped volume and assign it to copy of a virtual machine. In the next section we will show how user can do it.

To get access to you previously backed up data user have to make a clone of appropriate snapshot stored on Backup node:

1. Go to Backup node, and in Storage menu find Snapshots tab.

2. Select zvol that you want to restore by pressing Select resource and finding appropriate volume. This action will list you all available snapshots.

3. Select snapshot that contains most actual version of storage that you are interested in and press Clone.

4. Name new clone in a way that will resemble format v_vm-<vm id>-disk-<disk id>.

• vm id is a virtual machine identifier that is unique id inside proxmox
• disk id is unique disk identifier if disk associated with virtual machine 100. Where disk id is number that is expected to be unique respective to virtual machine with id 100. So that couple In out case it is v_vm-100-disk-1.

5. If everything is OK you will see new zvol in iSCSI targets tab of Storage menu.

6. Also volume v_vm-100-disk-1 will appear in VM Disks tab of storage pool jdss-Backup-node-Pool-0 in Proxmox.

Once volume is restored user can create temporary virtual machine.

To do it go to /etc/pve/qemu-server/

root@proxmox-demo:~# cd /etc/pve/qemu-server/

And duplicate config file for virtual machine that you want to restore. In our case it is virtual machine with id 100. Make sure that copy of original virtual machine config has unique id number. In this case we use 101 as it is next free vm id

root@proxmox-demo:/etc/pve/qemu-server# ls
100.conf
root@proxmox-demo:/etc/pve/qemu-server# cp ./100.conf ./101.conf

Open virtual machine config in your favorite editor

boot: order=ide0;ide2;net0
cores: 4
cpu: qemu64
ide0: jdss-Production-node-Pool-0:vm-100-disk-0,size=32G
us.iso,media=cdrom,size=5420408K
kvm: 0
machine: pc-i440fx-7.1
memory: 8192
meta: creation-qemu=7.1.0,ctime=1670424775
name: windows-demo
net0: e1000=3E:7B:70:D7:D7:E9,bridge=vmbr1,firewall=1
numa: 0
ostype: win10
scsihw: virtio-scsi-single
smbios1: uuid=7e718afc-0e17-4440-8432-0205b113d6ca
sockets: 1
vmgenid: e6ce6195-974a-4349-832f-6e7d4b115665

And modify line responsible for storage information. In this case it is ide0 record.

ide0: jdss-Production-node-Pool-0:vm-100-disk-0,size=32G

Make it point out to clone created in previous step by specifying Backup node storage pool and proper volume name. In this case it is: jdss-Vackup-node-Pool-0:vm-100-disk-1

So the line will look like:

ide0: jdss-Backup-node-Pool-0:vm-100-disk-1,size=32G

Also operator might want to change name field, so it would be more convenient to navigate.

name: windows-demo

change to

name: windows-demo-backup

So that final config will look like:

boot: order=ide0;ide2;net0
cores: 4
cpu: qemu64
ide0: jdss-Backup-node-Pool-0:vm-100-disk-1,size=32G
us.iso,media=cdrom,size=5420408K
kvm: 0
machine: pc-i440fx-7.1
memory: 8192
meta: creation-qemu=7.1.0,ctime=1670424775
name: windows-demo-backup
net0: e1000=3E:7B:70:D7:D7:E9,bridge=vmbr1,firewall=1
numa: 0
ostype: win10
scsihw: virtio-scsi-single
smbios1: uuid=7e718afc-0e17-4440-8432-0205b113d6ca
sockets: 1
vmgenid: e6ce6195-974a-4349-832f-6e7d4b115665

Once file is written virtual machine will appear in proxmox:

Now user have fully functioning virtual machine and can run it to make sure that all expected data is present. But we do not recommend using this virtual machine as production as its storage is located on a Backup node. There fore next best move would be to migrate this virtual machine to Production node

To do it operator have to go to proxmox and find virtual machine that was created from backup, in this case it is 101(window-demo-backup). Select disk jdss-Backup-node-Pool-0:vm-100-disk-1 in Hardware menu, and press Disk Action, there you will find Move Storage

Select resource pool associated with Production node, in this case it isjdss-Production-node-Pool-0 and start transaction.

Once it is done virtual machine is back to Production node and you can continue regular operation. Please notice that proxmox will rename volume. In particular case it would be named vm-101-disk-0.

Also we would recommend to set up backup task again. As this volume is treated as new one and no backup tasks is applied to it.