I now run HA OS in a VM. But I’m thinking to setup another pc with Proxmox and HA OS in a VM as backup. Any ideas, tips or tricks or links to use the best installation manual? My first vm was an installation done with community scripts and the second which I run now has been done with Proxmox helper scripts.
not using Proxmox myself, but you could check out Storage Replication - Proxmox VE
Just keep in mind that replication does not equal backup.
I have a Proxmox cluster with High Availability enabled using replication. The key to this is the ZFS file system. It also allows snapshots, which are great when updating and gives you the opportunity to roll back if an update causes problems.
My production HA instance will migrate between nodes on the cluster keeping it’s IP address. Note there is downtime during a user initiated migration (I don’t have the architecture or network for memory content migration) but if the migration is due to a failure there is downtime anyway.
The node restarts on the new cluster with up to a 2 hour gap in history if a hardware failure occurs (I replicate only every 2 hours) which is fine with me. You can replicate more often if desired and your infrastructure supports it.
The majority of my IOT devices/sensors are IP based so no need to account for them in this architecture. I have a large Z-wave networks but my Z-wave controller is a HomeSeer Z-net which is IP based on a RPi so the HA instance will reconnect to it fine when a migration occurs. Same process from my Zigbee controller which is plugged into the same RPi.
Consider how any external hardware to you HA VM can be reconnected. Serial to IP is a good strategy for high availability. Anything you can move to IP will support this HA strategy.
Good luck.
How can I check if I have this active too? And how do you configure the HA migration between nodes on the cluster with same IP?
I use the SLZB zigbee coördinator with SLZB repeaters to control the Zigbee network.
OK I started with ZHA and still have some things on this and this uses an USB port in the Proxmox server but most of the stuff is controlled in ZMQTT.
I use ZHA for my Zigbee software and it will accept the ser2net TCP connection directly:
Z-wave JS UI will do the same.
You mean High Availability on your Proxmox cluster? For Proxmox 8.x:
Make sure you have a Cluster:
Proxmox UI > Datacenter > Cluster:
Make sure the VM has replication enabled:
Datacenter > Server > VM > Replication:
Configure High Availability for the VM:
Datacenter > HA:
At that point you can manually migrate on the fly or if the server fails the VM will migrate automatically:
I’m using a 2.5Gb dedicated migration network and things work pretty smoothly. VM replication takes about 10 seconds for Home Assistant.
Thank you so much for this tutorial but I still have some questions if I may ask.
I do not have this setup yet.
You have a seperate Zigbee software server running I guess?
I use Zigbee2MQTT in my HA OS VM with SLZB-06P7 coördinator and repeaters.
I do not have the clusters only 1 node called proxmoxserver
I run ZHA as the Zigbee software that runs in Home Assistant core.
One Proxmox node equals no high availability. You’d have to have more Proxmox nodes to support a cluster and then High Availability.
Back to your original question, you can use the same helper scripts to build a second Proxmox server, then join it into a cluster with your first server.
Both need to have ZFS as the VM file system, however. You don’t say if you are using ZFS.
If your run your SLZB over ethernet instead of connected to the PVE with USB, you will not have to move that connection if/when you need to move to the second PVE.
You also do not need to load HAOS onto the second PVE now. Make sure you take backups of your HAOS VM. Then if there is a failure, you can bring the VM up on the second PVE with a restore.
I do not use High Availability. But do have my PVEs clustered. I have 15 VMs/LXCs installed across a cluster of three PVE machines. I can migrate the guests between the nodes as needed to take a node offline to do any upgrades or whatever.
I run PBS (Proxmox Backup Server) in a PVE VM to handle all the backups to my Synology NAS. I can restore those to any of the nodes if/when needed. One PBS VM handles all backups across the cluster.
Excuse me for not be clear.
Yes I’m using ZFS file system.
So it is best to have a 2nd hardware server with Proxmox and the same VM for HA but how do you start the cluster because if I run 2 VM’s with HA it will brick the communication with all my zigbee and other things at home? or do you have to add the 2nd Proxmox with VM HA after you run the first one?
Also very interesting your setup. I’m just not very familar with the given that if you run an extra PVE machine and how will the system know which VM is the master at that time because if you run more HA’s there will be issues with the zigbee network, correct? The most of the zigbee modules are connected through SLZB over ethernet LAN and I also have a few ZHA modules which now are connected through a Sonoff USB dongle on my PVE server and I have a BT dongle which is used to connect to my Solar Sunny boy to read the values. I then should have those USB dongles put into the other machine when the master fails?
Usually when you replicate a virtual machine only one of the vms are powered on. It replicates on the storage level. USB dongles will have to be moved from one host to another in the event of a failover.
Only one PVE machine has the VM running then? But the data is replicated between the several PVE machines with cluster?
High availability: Usually you run the virtual machine on storage that is accessible by multiple hosts, so that in the event of a hardware failure on host A you can reboot the machine on host B
Replication means you replicate (two or more copies) the virtual machine between two different storage units, so that you also can survive a failure on the storage.
Both involves only one running instance of the virtual machine. If you replicate you have a cold copy of the vm, with high availability it’s only one virtual machine.
Is it useful to install Proxmox VE 9 on new machines to run VM HA or stay with PVE 8.4 ?
And upgrading other machines too from 8.4 to PVE 9 for having the same cluster environment?
I would stay on 8.4 for now. As mentioned above there is only one VM active at a time. The VM disks are replicated between nodes on the cluster. No issue with multiple zigbee controller connections as there is only one connection at a time. No USB connections as mentioned above however, as I am using TCP to the zigbee coordinator. The coordinator connection simply moves to the new VM when the migration occurs.
What did you configure on your NAS to have it good and quick accessible please? Through Samba or NFS ?
I don’t replicate the data. That is what you want to do for High Availability.
I use backups. The VMs/LXCs are backed up nightly.
Actually, the way I am using it, you don’t even need a cluster. You could just restore to the other PVE if needed.
The benefit of clustering for me is that I can control all three PVEs with a single login. If not clustered, you’d need to log in to each PVE to do something to it.
I can also migrate guests between the PVEs since they are clustered. Saves having to backup and restore. It does it for you.
Initially, I just had an NFS mount to the NAS.
Once I added PBS, I setup iSCSI on the NAS following this article:
I use PBS to backup all guests except the PBS VM. It cannot backup itself. I use the PVE backup for that one.
Thank you all for the support, which installation method should I use for installing Proxmoxserver with HA OS in VM ?
I used the community script.
Oh, I see Derek’s blog is the whole thing from installing Proxmox bare metal to using the helper scripts. That is a good resource.








