VM HA OS loses connection in proxmox

Hi, I am currently running my HA on proxmox with a VM. Until the core version of June everything has been fine but if I try to update to the core version of July, HA restarts fine and the VM starts to work fine and after a few minutes, in the proxmox interface, the assigned ip disappears to the VM and exits gest agent is not running, so I can’t access HA even locally. Sometimes, after a few minutes, it recovers the IP again and then loses it again. If I ping from Windows, the VM ip always responds. I also have two CTs in proxmox, with them I have no problems. If I return the HA core to the June version, the problems are over. To say that I have fully updated proxmox and the NUC on which it works connected by cable and with a fixed IP, both the NUC and the VM and the CTs. Any idea what can happen? I have tried things but my knowledge is not very great and I don’t know what to do anymore. Sorry for my English, I’m using a translator. Thanks in advance.

In Proxmox under Datacenter click on the node (default is pve) and then on the right sidebar click on syslog (under system heading). Please provide the system log so we can assist you.

Thanks for your answer. I have been able to see the error that the proxmox system log gives when the VM loses the connection. I don’t understand too much but I think it’s a problem with my network. The error is as follows:
Aug 01 15:25:49 proxmox pvedaemon[1090]: VM 103 qmp command failed - VM 103 qmp command 'guest-ping' failed - got timeout
With the June HA core release, I don’t have this problem, the VM doesn’t lose connection.
If I put the hardware dedicated to HA, could it solve the problem?

Home Assistant in the Proxmox VM is a HAOS installation or possibly supervised or core? It seems you are running core, not certain why. If you want a reliable, trouble free HA implementation with Proxmox VE your VM should be based on HAOS.

In case you are interested please read my comprehensive guides below:

Windows is relevant if you want a permanent infrastructure management station on your network based on Windows. You can also use it as a media server by just configuring everything through Windows Media Player.

I have more published guides on HA if you have the time and/or the interest.

Thanks for the help. My installation on proxmox is a HAOS VM. Since I started with HA I have always had HAOS. When I talk about the core version, I mean the updates that HA has, for example the latest is 2022.7.7. The last version that does not give me problems is 2022.6.7. I am going to read the information that you have given me although with my limited knowledge I do not know if I will be able to solve the problem that I have.

Can you send the configuration of your HAOS VM. Under Hardware and Options. Focus on Network, CPU, Memory and Disk config. Screenshots of both Hardware and Options required. Also screenshot of your HA network IPv4 config.

Hello, I attach images of the configuration, probably more images are missing. HA OS DNS is the ip of pihole. IPs are static.
If more images are needed, just tell me.

From the images provided no smoking gun detected. Can you please send a screenshot of the net config of your pve node: pve → System → Network.

Please check your DHCP server is not serving DHCP addresses above or below You might have a DHCP client allocated the same IP address as your HAOS instance. You DHCP server is on the gateway/router or Pi-hole? A screenshot of your DHCP server settings would be useful.

If above do not solve issue try the following.

Shutdown HAOS VM, clone HAOS VM and start the clone. Assign a new IP address to the HAOS clone e.g. (something outside the DHCP address reservation pool). If issues persist I suggest you create a new HAOS VM, and see it the issue persists. On every new HAOS VM you create make certain you are using a valid static IP address, something that will not conflict with a DHCP address.

I enclose an image of the network configuration of my proxmox and also of the DHCP of my router. Pi hole I only have it set as DNS in my router, nothing more. I still need to try the change of ip to a range outside of DHCP. I’ll explain my network configuration for what it’s worth: Vodafone router with Wi-Fi disconnected, DMZ enabled and bridge mode disabled. From the wired Vodafone router to a tp link mesh wifi with two nodes, the cable goes to the first node. The NUC where I have proxmox is connected to the first node also by cable. I’ve been with HA for 2 years without any problem like now and since November with proxmox and without problems too. I don’t know why when moving to HA core version 2022.7, the error occurs.

I suggest you change the IPv4 address of HAOS to something below 100; e.g. or below. Right now your static HAOS IP address is within the DHCP address pool, which from a networking perspective should not be.

Also change your Proxmox linux bridge IPv4 address to be outside the DHCP address pool. anything below 100 and above 250.

The issue is that because something works does not mean it is not wrong. You network setup require you to follow one rule only. ANy static IP address should be outside the DHCP address pool scope and in your case any static IP address should be below < 100 and > 250.

I have changed the DHCP ip range on the router so that all my devices have their ip outside the DHCP range. The problem persists, 4-5 minutes after starting the HAOS VM it loses the connection, the ip of the VM summary in the proxmox interface disappears and the CPU and RAM usage rise to more than 100%. After 5 minutes everything returns to normal and after 10 minutes, it loses the connection again and it has stayed that way.

2022.8.0 will be out tomorrow, maybe that will work for you? :crossed_fingers:

I’ll give it a try, hopefully it works. If not, I am considering installing HAOS without proxmox with the NUC dedicated to it. BTW I used your script to create the new VM for testing.

1 Like

If that version works good, I can’t find a reason why 2022.7 wouldn’t
Did you try the PVE Edge Kernel?

That’s right, version 2022.6 works without any problem. As soon as I update to version 2022.7, the VM has the behavior described in previous posts. If I go back to version 2022.6, it works fine again. Regarding the PVE Edge Kernel, it is currently the one I have installed. In summary version core 2022.7, version operating system 8.4 and edge kernel, the VM has the wrong behavior. On the other hand, if the core version is 2022.6, operating system version 8.2 and with edge kernel or Edge no kernel (I don’t know what to call it so that it is understood), the VM works normally. If in Windows I use the CMD and I put ping -t ip VM HAOS, it responds to the ping and continues to respond even if the erroneous behavior of the VM occurs.

So, it just doesn’t like 2022.7
Hopefully 2022.8 will set things straight :crossed_fingers:

I hope so and you’re right. I will keep you informed.

1 Like

Can you create a new HAOS VM and see if the issue manifests. I assume your Proxmox is up to date. If not, update Proxmox, reboot, and see if problem persists.

There is a new kernel update in Proxmox in the no-Subscription stream. You can try this update prior to trying a new HAOS VM.