I had HASSIO running on a RPi3B before CoVID but have not been monitoring it. I now start up the RPi and get the graphical interface, but HA does not launch (not accessible via IP:8123 )
It is installed in /usr/share/hassio/homeassistant but I cannot find or get a response from any CLI commands hassio, hass or ha.
I used Get-apt to upgrade the OS but can’t get into HA to do any patching / updates etc.
Can anyone point me at a good script / procedure to recover my HA instance or do I have to start from scratch? It was running well on a RPi3B with connected SSD, and the latter took a lot of effort to get working so would rather not start from a brand new image.
[Also Home Assistant Blue looked like an interesting development, but did that expire when stocks ran out or has that project moved on and called something different?]
Hi,
The answer rather depends on cost / benefit and your skill level.
Setting up an RPi with external SSD suggests you have at least some Linux sys admin knowledge.
Having done so, there is some time invested in the existing configuration.
If you can get the existing install to the CLI, the key question is the state of the data partition used for HASS (/config which is a ext4 filesystem). That it’s not starting suggests there should be error messages shouting loudly about fs errors. The failure mode is likely to be one partition full / corrupt / too many writes. SSD / uSD really don’t fail as often as they used to, although USB3 SSD are a lot faster on a RPi4 / 400.
Were I going for full data forensics, I’d create a copy of the disk / partition as an ISO, and use loop-back mounts to work on a copy - leaving the original untouched. ddrescue is a useful start as it copies what it can, ignoring errors. You can then try fsck.ext4 or even grep as a last resort to get text YAML back. There’s lots of tools, but they all take time to learn and use.
Use the original RPi with a fresh RPiOS uSD - that way you’re less likely to make a mess of your main machine’s data partition! Linux recovery on Linux is a lot easier (a grey beard sys admin / dba / architect would say that of course).
Another approach is to build a fresh HASS install uSD, then remove it and use that disk as a test to better understand what the corrupt install should look like.
I’m not going to mention backups - intruding into private grief is not good (Using SCP on a web shell to copy the major files to another machine is a good start).
this post!