I spend my winters (5mos) in the desert which means my HA host at home must be able to reliably deal with power interruptions.
The worst thing you can do to a modern OS is power it off without a proper shutdown. Note also that power is often very ‘dirty’ as it is being restored and can overwhelm a power supply’s ability to provide clean power. Homeassistant typically does not have to ‘work’ when the power is out, but it does need to continue to function after the power outage without intervention. This is why a UPS is so important.
A UPS is not just there to keep the HA host up during an outage. It will signal the HA host to do a proper shutdown if the UPS is running out of battery power. This means you must have a service running on the HA host that communicates with the UPS. For me the easiest way to do this is to use a laptop to host HA. Most any generic Linux will ‘just work’ You then need to run HAOS in a container/VM. You can buy an old/refurbished Windoze laptop for $1-200 that is more than up to the task. The laptop will typically stay up quite some some time, but if you want to handle longer power outages, you can add a UPS. The downside of a laptop is that if it shuts down, you will need someone to press the power button.
Good point. I bought a small UPS for my RPi which I’ve found works very well. I didn’t go the extra mile and include a sensor with shutdown and restart, because around here most outages are brief and longer ones are rare.
If I were gone for months, I think I’d want a more robust UPS strategy, and probably one of those devices which senses when the network is down and reboots the router and/or cable modem or whatever you use for WAN connectivity.
HA (home assistant) can function in a HA (high availability) setup and can have tier 4 datacentre uptime with comparatively low budget. Problem is your house services have not been designed for high availability, but if you have deep pockets any home can be designed and operated as a tier 4 datacentre, with virtually no downtime whatsoever.
Is it possible to run HA on AWS now? I finally managed to fire up an EC2 instance (Debian 12, downgraded AppArmor to v2.13.x), but it is Unsupported and Unhealthy: network_manager and privileged issues.
I’m quite beginner in Linux, is it possible at all to make it Supported and Healthy on AWS?
Eg. network_manager is installed, but there are no physical interfaces.
The only issue I have here is “what if there’s catastrophic hardware failure?” Is there a way to have backups somewhere? The rationale for the cloud (in your own self hosted instance) tbh is mostly if you have hardware failure, in the home it surely would be more involved? In the cloud it’s like “Eh someone else will sort it and everything will just work”