I installed HA a while ago (a year or two ago) on bare metal and it has been working pretty nicely since. However, it died a couple of days ago and after a hard reset of the computer it came back online. I checked the logs and I see a ton of errors like this:
homeassistant kernel: I/O error, dev sda, sector 49109360 op 0x0:(READ) flags 0x0 phys_seg 1 prio class
This tells me that the ssd is probably about dead, but I wanted to use smartctl to check it out. Does smartctl really not exist for HAOS? I have been searching all night trying to find a way to install it, but it doesn’t seem to exist for HAOS. That cant be right? I’m surely not the first person that wanted to check their drives.
[home_assistant - /dev/sda - sat - Samsung SSD 870 EVO 1TB]
Last Updated on March 26, 2025 - 17:50
Status Temperature Capacity Powered On
Failed 30°C 931.5 GiB 3 years
Thank you! You were spot on!! After turning off protection mode for the addon, I was able to get it running and unfortunately it proves that the drive is “mostly dead”. Still under warranty, so I guess I’ll have to replace it with something else while I wait for them to send me another drive.
Thinking of installing HA as a supervised installation this time so I can more easily use standard Linux packages without installing addons. Seems to be about as much work as installing it as HAOS with the added benefit of being able to run native Linux commands.
Just FYI, you are not permitted to install extra packages in a supervised install. The only install methods you can do that to are a core installation or a VM.
Personally I’d go VM. There are discussions about dropping support for core and supervised installs as they take up a disproportionate amount of support compared to their userbase. Nothing set in stone, just a discussion at this stage.
Oh wow, I wasn’t aware of that. Thanks for pointing that out as I might need to change direction then. Do you happen the know the rationale for that decision? I know that HA is trying to cater to everyone and maybe is trying to take the Apple approach and make it just “work”, but I can’t be the only one that wants to run HA directly on hardware for the best performance and ease of use.
This thread alone is kind of proof that we need the flexibility to install regular Linux applications to troubleshoot and administer our installations. This situation was a hardware failure and thankfully someone was kind enough to create a docker container that was compatible with home assistant, but if I had the ability to install packages, it would have taken 10 seconds to install smartctl and do the maintenance myself without having to worry about installing another docker container and configuring it (and that is assuming that the needed functionality was already created by someone other than HA).
Maybe I am looking at this the wrong way and you could help me figure out the best way forward? My goals are to never have a hardware failure like this without being able to diagnose and fix the error easily using standard commands. According to the HA Installation page, if I want to get all the features of HA I need to run it as either HA OS (which is what I am currently running) or Supervised. Container and Core loses out on Addons/easy updates so those aren’t installation methods I would consider. I have a 3050 SFF PC that is dedicated to HA, so my options would be:
Install Debian on bare metal and install HAOS as a supervised installation. This seems ideal because I would have full control of the hardware and can install packages (well I thought I could).
Install Proxmox on bare metal and run HA as a VM. This would be do-able, but I would have to pass through all the hardware devices and the performance would suffer.
Install HA on bare metal like I have now with the same limitations.
Is that correct? I would like to do option 1 after this experience, but am confused about nothing being able to install other packages.