Corrupt Archive HAOS 17.1

Installation
1

Hi all, very very new user to HA, as in I started using it a few weeks ago. Yesterday I upgraded HA to the new incremental update and thats where its all gone wrong.

I am using a HP Prodesk 400 G5 Mini, and the system just locked up on the update, after a lot of prodding I decided to just reinstall (I had a backup that I ran before trying to update, Only smart thing I did yesterday I think).

So I use Ubuntu on a flash drive in the HP download “haos_generic-x86-64-17.1.img” and using Disk I install it, Disk not happy saying

“Error reading 1048576 bytes from offset 1129316352: Invalid compressed data (g-io-error-quark,35)”

Tried a few things using Gemini to solve it, no joy, try and install 17.0, great that goes on all okay. Boot into HAOS and it tried to install 17.1 and it just doesnt want to play. I get this message in red:

Screenshot_20260308_081043_Edge
Screenshot_20260308_081043_Edge1080×2105 208 KB

This morning I thought lets see if I can try something else, so i go grab an old SSD and tried to put haos_generic-x86-64-17.1.img on that using balenaEtcher. Also no joy, just a message saying “something went wrong. If its compressed image, please check the archive is not corrupted.”

My conclusion is either I have 2 corrupted Hard drives, im an idiot or something is up with the 17.1 file. I dont think its 1, high chance is 2 but could be 3. Any help you can offer a newbie would be greatly received, I spent about 9 hours on this yesterday and out of idea, even how I stop it trying to update to 17.1 to 17.0 could help, at least then I can use the system again.

2

Just to add I have tried the 2 install options advised in this post Generic x86-64 - Home Assistant are the ones I tried and it wouldnt take 17.1

3

I suggest you “start” here, install the working Ver. and install your Apps, integrations etc.
you could post which CORE, SU and Front-end versions you got up and running

Be aware that Many People have up graded to both OS17.1 And CORE 2026.3 etc.
So far your issues might be related to other Issues than a Corrupt HAOS archive in Github, so more likely something in “Your end”

4

I am going on option 3 atm (Problem between keyboard and machine). But i have have repeated the same stapes in the guide multiple times, 17.0 goes on, 17.1 fails to burn onto the drive.

How do I stop 17.0 upgrading to 17.1 as soon as HA starts as thats where I am stuck, i turn it on with a fresh build, console starts up, I go to http://homeassistant.local:8123/ and 1st thing it wants to do is update to 17.1, I have no mechanism to stop it.

5

Right,You got me there :wink: , i honestly have no idea
I have never heard of this “issue”
What you could try is when 17.0 console shows up ( Don’t access Url in browser )
In the console:
Type ha os update … And see it this work , Give it time

But before you could also type " ha core info " and " ha su info " so you have these version numbers “noted”
Could be so that you need to update CORE etc to February first

6

Well I prodded and poked it with a big stick, unplugged the router from the internet to see what happens and it didnt like it, HA wants to update to the latest version on set up, and if it has no internet connection it will sit and wait.
Plugged it back in, it started to do its thing and it fell over, however it failed leaving me with 17.0 which was useable! Quick update to supervisor to get it to the version I had my back up in, rean the back up and back up and running.

I have no idea what the issue is, leaning towards hardware at the moment, but its working, I have no need to update it right now so I will just leave it be.

1 Like
7

You could try in CLI

ha core update 2026.2.0
Then
os update 17.1 ( I think you need core 2026.2.0 for os17.1 )

After this then monitor your logfiles regularly to see if you gets repeated daily errors
This way you will get an idea how your system works/feels, and are better prepared for Troubleshooting, while knowing how your Logs looks like daily

8

Supervisor will try to load 3 times, and then fall back to the previous.
There are also cli commands to specifically pick versions, switch slots, all supervisor related.

1 Like
9

Just wanted to come back with an update, been waiting on a new stick of RAM to arrive after doing a test using memtest86, the RAM was totally FUBAR. The test self aborted it was that bad so that explains the issue, because the system stores it in the RAM before sticking it onto the SDD and the RAM was corrupted it was failing the CRC.
So looks like a very rare edge case ive had but one I thought worth coming back to and updating in case anyone else has it come up!

2 Likes