I am running HA OS in a VM on Proxmox VE, for almost 2 years without issues. Now I wanted to maintain the PVE host hardware and therefore migrate the VM to another PVE node.
However, HA does not run properly on the other node.
First, I made a backup of the VM and restored on the second node. Here, the the apps/add-on did not seem to start or were not reachable for HA and from other host in the network. HA was very laggy too.
Second, I actually migrated the VM to the other node in the cluster with the dedicated PVE function. Here, the frontend was not reachable anymore after some minutes, but the Observatory URL did show that HA was healthy. Apps did not seem to have spun up, as e.g. NPM service was not available.
The VM uses USB devices, which I have mapped to the second node before migration.
Migration of another VM running Ubuntu succeded at first try without issues.
The hosts are Mini-PCs with N100 and N150 CPUs respectively, both with 16GB RAM and 512 GB SSD.
I made a new HA VM on a new Proxmox Node and loaded the backup from the old HA instance.
At first everything looked great, but after some time th web UI is unresponsive and very slow.
After a VM restart the UI wasn’t reachable anymore at all.
So I went back to the snapshot before loading the backup and tried it again.
Still very slow and unresponsive.
What I noticed, is that there is no logfile. Just the .log.fault
What’ also weird, is, that i was asked to enter the encryption key.
That was with v2026.3.0
Today there is a new VM with 2026.3.1 that doesn’t ask for the encryption key.
When restoring backups, then be aware that the network might have lingering entries having around, like mDNS bound to specific IPs or ARPtables in the network gear that needs to update itself.
Running two instances of HA at the same time makes it even worse, if these things have not been handled, since the servers will then fight over the mDNS names and maybe even the IP addresses too.
Make sure that IPs and the server name are different between the servers.
How did you manage the storage migration? You mentioned dealing with USB, but the new host has to have copies of the same disks, unless you are using some type of SAN, which it seems like you aren’t.
Since you’re on Proxmox you have web access to the console. Login to the console and check the status of docker containers. At the HA> prompt enter login, then docker ps. Is the Home Assistant docker container running? Likely not. Try restarting in safe mode in the CLI.
The network might still need to catch up, especially the mDNS.
Did you also give it the new server a new hostname or did you shutdown the old server, so it is not present on the network anymore?
I did not run two instances of HA at the same time. If migrating, this is not even possible as the VM on the original node is deleted. If restoring a backup on the second node, for sure, the original VM needs to stay stopped.
Running under different IP or hostname is no option, as so many other devices are connected to HA.
Network caches might play a role, but accessing HA via ssh or ping was no problem.
I also experienced that the logfile was not updated on the file system, although in the GUI it was present, if the GUI was responsive. Moreover, apps did not start as some of their internal services ran into timeouts. I have a smart meter adapter that sends updates every second. Usually no problem, but here it missed values for up to 18mins.
CPU load was very high, close to 100% all time.
That brought me to the solution: When setting up the new Mini-PC with N150 CPU, I set its performance mode to “normal” in the BIOS. According to the manual this is for permanent operation with reduced demands - fit for HA I believed. Hence, I switched back to “Performance Mode” - and everything works fine.
No timeouts, UI is responsive as usual, no missed values. Finally, migration succeded. However, longtime CPU load is about 15% on the new N150 CPU, where it was at 8% on the N100.
It is not always possible for us to determine what the description mean in reality.
Some migrate through the Proxmox migrate feature and others migrate by making a HA backup and restoring that on another device.
We just try to cover as many bases as possible.
A couple of updates back the log file in the file system was disabled, so the one you see might be an old one.
If you want to have a log file in the file system, then search the forum for a solution. Someone made an integration or addon that restores the log file.
I also found the solution.
I also saw that a lot of integrations didn’t start up.
For me it was a Modbus integration.
I had a onboard RS485 in the old IPC.
The new one has an USB RS485.
I changed the yaml file from ttyS0 to ttyUSB0 as shown in the hardware.
I had the USB working on an Raspi so I am pretty sure the dongle was working.
Never the less, once I deleted the Modbus part from the config everything started just fine.
Since the Update to 2026.3.1 was basically in my migration window I rolled back to 2026.3.0 and pasted the modbus config again.
Still running fine.
So I guess something happened with the update to 2026.3.1