In the supervisor, I see “You are currently running version 4.16”. But when I click “Update”, the dialog box says “Are you sure you want to update Operating System to version null?”
Clicking “Update” gives a new dialog box: “Update Failed - Unknown error, see supervisor”
The system logs show
20-11-28 22:58:07 INFO (MainThread) [supervisor.hassos] Fetch OTA update from https://github.com/home-assistant/operating-system/releases/download/None/hassos_ova-None.raucb
This is a clean (almost entirely) install, I just spun up the qcow2 image and added Samba and Appdaemon from the Supervisor add-on store before trying the update.
Leaving it overnight, it now asks “Are you sure you want to update Operating System to version 4.17?”. So maybe it just needed more than a few hours to get set up?
However, I can see that the latest stable release is 7.18, so it’s a bit late to the party. Is that a general thing for the HassOS? We’ll always be a release or so behind? Not a huge issue for me (Home Assistant version is 0.118.4, so up to date), just wondering.
I ran Home Assistant in a VM for nearly two years and never waited hours to update anything. what type of resources did you provide to the VM, how big is the host?
Right now I’m just testing on an old mini PC. I wanted to make sure I knew how to set up oVirt and install/run VMs before messing with my current setup (Ubuntu Server -> Docker) running on a proper server, since I’ve never done more virtualization than a Windows in virtualbox/VMware on a MacBook.
As of now, it has been allocated 2 vCPUs, 2 GB of RAM, and running off a USB HDD (on a Fujitsu Esprimo q520 with an i3-4350T and 8 GB of RAM).
Ok I see, the USB hdd is not an ideal situation. I ran Hassio VMs with 2 vCPU and 2 or later 4gb ram successfully for quite a while. My old setup was an i5 NUC with an M.2 drive running XenServer (no USB pass through) which i replaced with an i7 NUC and M.2 Drive running ProxMox. The USB bus is going to be your biggest issue due to the additional overhead the hypervisor has to run the VMs on the USB HDD. Keep in mind that USB throughput standards are not what I’m talking about. Latency on USB is quite a bit higher than a SATA connection or an M.2 interface. How the hypervisor communicates to the VM for transient resource allocation (memory and compute) of the VM is running on the same latent data bus as the application transactions which need to be real time. If you move the VM to a native drive spinning disk or SSD, you will see real performance changes. Additionally HassOS is more than just Linux, it is a microservices environment. You are running a container environment in a VM on a constrained path to all the physical resources. The i3 and the 8gb ram are more than enough to run the system. If it were me, I would shed the hypervisor and image HassOS to a Local Disk. In my current setup I moved back to the i5 and imaged the drive with HassOS directly gaining my best performance to date. My final thought is, if you need to stay on a VM make sure you have intel VT-x support otherwise the hypervisor will always be slowish…
I hope this helps you in some way, My experiences were gained based on my success and failures both my hobbies and in my professional life. Good Luck!!
Great, thanks for the input! But no worries, as I said, I’m just using this setup to test that I can set it up so I don’t end up having days of HASS downtime once I start.
As soon as I’m confident I can do that (and have a spare moment) I’ll move centOS/oVirt to an old IBM X3650 server (the one that’s currently running HASS, Nginx, website, DokuWiki, Mosquitto, etc.). The current PC will be used for something else, so the server is what I have available for HassOS (and a few other things; 96 GB RAM, 2 x Xeon E5520 is a bit overkill for only HassOS).
Once I move, there will be 136 GB + 2.7 TB RAID1/5 arrays (though spinning disks, but haven’t noticed any large-enough-to-care-about latencies yet), so the only USB I’ll need is the ConBee II and AeoTec ZigBee/Z-wave sticks (and perhaps a backup USB HDD just in case).
I’ve been fine running the Docker image for years, but I see the way the wind’s been blowing the last year or so, and thought I should finally convert to using the supervisor (in a “supported” install) to avoid being more or less forced at some less convenient point in the future.
sound like a plan, you should consider using the the HassOS image for your VM, I went from Hassio on docker CE/Debian all custom built as that was the way we all did it in the past to HassOS and I have been very happy with the prebuilt images. I am running my Dev instance in that fashion and its rock solid.
That sounds good! Because I am using HassOS (qcow2 image). At least that’s what I’m testing, and planning to set up. Getting a full (supervisor) supported install was the entire reason for me to change my setup (move away from Hass Core docker container).
Honestly, I think I’ve just been curmudgeonly sticking to Hass Core container a lot longer than perhaps I should, since I liked the full control of my base OS.
