Disaster Recovery Planning

Have you ever thought about what you would do if Home Assistant stopped working due to a hardware failure (sensor, PC, power, etc.) or other problems?

Failure of HA

Backup & Restore Process

The first step is to have good backups. But backups are only good if you know they will recover your system. So, regularly test the restore process before you need it - you do not want to discover that your restore process fails when you need it. A good practice is to keep your backups separate from the hardware running Home Assistant. Methods to do this automatically can be found here: Copying your backups to another location.

If you can use add-ons, Samba Backup is excellent - backups can be scheduled and can automatically copied elsewhere (to a NAS for example).

In addition, it’s not a bad idea to take copies of the yaml files in your config folder from time to time - they’re just text files. If you do this, it means you can reinstate things - a single automation, say - without doing a restore of the whole system.

Again, there is an excellent add-on for this. Samba Share makes it possible to access HA files from other machines on the network.

Add-ons are only available if you have a Home Assistant Operating System or Home Assistant Supervised installation.

What to Do While HA is Down

How will you and your family manage without HA? Will light switches still operate without HA? Will critical sensors (fire, intrusion, temperature, etc.) still alarm without HA? These types of questions should be considered when automating anything.

When HA Fails & the Recovery Process Fails.

Stuff happens…. Being creative, flexible, and determined may get the results you need. Don’t despair - search this forum and post questions.

Failure of individual devices

Redundancy

If you have a critical device - a water leak sensor, perhaps - consider having two with different protocols, one Zigbee and one wi-fi, for example. The chances of them both failing at the same time are very small.


The Home Assistant Cookbook - Index.

3 Likes

I speak from experience (a frozen sump pump in February). Critical alarms should ALWAYS have redundant fallbacks. And test them monthly. In my case, my Alexa connection requires, at random times, a refresh of the cookies. Without it I don’t hear the TTS announcement that the basement is flooding. Today I have three flood alarms- two of them have no online presence.

You also forgot the ultimate disaster plan- what your spouse will do with your installation when you are gone.

2 Likes

Preferably offline and audible or visual.

Get up and turn light on myself. With the exception of 2 lights, everything may be controlled manually. This idea falls in the category of “what happens if power goes out or server breaks”.

I have backup hardware and can be back up in under an hour if I wanted.
If power goes out, I will just get candles and wait.

1 Like

The ultimate horror, I will have to use wall switches and operate the thermostat manually. That is, if both my primary and backup/test instance fail at the same time

1 Like

@stevemann I also had my sump pump mounting altered such that it was very easy to remove, and keep a spare nearby for a quick swap which I test every three months. I also actually have a spare water line leading into the pit which I can turn on to test the sump pump occassionally (if it has not been run in a while). Also, I have a generator (not whole house unfortunately) nearby which I can start up and some very expensive, long, hugely thick extension cords (so heavy duty they can’t even be easily coiled up) nearby to plug into the regigerator and sump pump from the generator. As you can probably tell, I had a near catastrophic situation and swore to my wife that would never, ever, ever, ever, ever happen again! Also, instead of waiting for a failure, I will swap out the old sump pump with a replacement and then buy a new spare, every 5-7 years as well. Also, I have not only yolink leak sensors nearby (cloud based -yuck) but also have installed Level Sense Pro. Then, I have a water sensor inside the sump drain pipe where it bends to go through the foundation wall - which reports to home assistant and also adds a line into a google sheet - every time the pipe has water flowing and when the water stops flowing. The reason why I have the yolink sensors instead of local only, is that it is tied into their ViruAlarm service which will wil have a live person call mt cell phone about any water leak detected, as well as call the authorities if I do not respond*. A bit of overkill, yes…

*Vituralarm, when it is working 1,000% flawlessly - I plan on also used with all the other Yolink Sensors (motion, door, window) as an alarm system.

For redundancy for the sensors I also have numerous Yolink AND Shelly (Motion BLU and Motion 2 models) sensors in some of the same rooms, where I need 100% accurate detection (bathroom for keeping the lights on while people are in the shower so they are not left in the dark, etc.). I have ot yet but will eventually replace some of the motions sensors (all of which are PIR) with the more sophisticated ones coming out (mmwave, radar, etc.)

This “hobby” gets expensive… Weather station tied into HA to remind me which widows to close when it starts raining, etc…

One problem I have is my RPI4 & SSD failing… I do have a backup but my spare RPI is a 5 not yet set up… sheesh! Also I have the entire thing set up so that light switches etc. will still work anyway so if the RPI dies It’s not a disaster -

Besides backups (I love HA and don’t want to go without), I do design my house to be HA-disaster proof. So my door will open when my smart lock is dead, my lights will operate without HA, my thermostat is pretty self sufficient, HA just makes it smarter. And indeed, it is something I plan for.

For some critical things an ESP can also be a solution, as it will run without HA. But it can be a nuissance that it reboots periodically if it can’t connect to HA. But that can be turned off.

2 Likes

Nice to see people who have planned and tested for this!

But how do we “spread the word” and get those who have not done any planning or testing to do so (or at least think about it)?

This would be a good topic for tips and tricks, and that topic would deserve a link in the

but I see that this post has already been added there. So the right questions are there. Let’s work on gathering tips.

My ultimate alarm is a float switch in the sump wired to an old police-type siren and a 12V car battery on a trickle charger. It’s guaranteed to get attention- even from the neighbors…

In my case, the exit pipe from the sump froze shut and the 15 year-old pump couldn’t handle running 100% of the time. Amazon wasn’t connecting my Alexa devices so no voice warning, and most embarrassing, the simple moisture detector batteries had also expired. It could have been a lot worse if my wife hadn’t asked “what’s chirping in the basement”? It was the moisture detector with dead batteries trying its best to scream.

I have a page in ui-minimalist where I display in one place on my phone the battey levels of every one of the 66 sensors I have in my house. I can easily see which ones are low and once every 6 months or so just saunter over and replace those -

My old Ring doorbell is wired but I have to turn off motion detection (used to turn on the front lights and also record in the middle of the night) until the battery recovers from time to time. I have a spare battery but it shouldn’t be necessary to either turn off recording or switch and charge the one not in use as the whole thing is powered by wires anyway! I will be replacing it with something that isn’t such a PITA - the motion that is sensed only makes it into HA half the time anyway - crappy device, API (and therefore integration for HA)…

Anyway, the above are about half of the the battery status’s but you get the idea (I could have something that would just remind me or whow me which batteries are low - but in case that fails or I miss the message - I figured its better to be able to see them all in one place).

I have HA connected to a smart plug that can be accessed outside of HA. When away from home I have been able to power off and on the HA os when it became uncontactable. Oh, and avoid updating HA while not at home. I have 2 VPNs to connect back to home and neither go through HA. Thermostats can also be accessed this way if HA still fails to come up.

Great idea @Spiro, I used to have that but my RPI is also running Weewx and other daemons and just pulling the power would corrupd the weewx database, etc.

About sensor failures: does someone have a simple way to use two (for instance temperature) sensors in automations?
I recently had a minor issue: the pool heater continued heating because the pool temperature sensor failed to communicate (WiFi failure).
Having a second sensor would be more than useful, provided the automation is able to handle them.

Not claiming this is the best solution, but it’s worked well enough for me.

I have a main temperature sensor and an alternate. If there’s been no change to the value of the main sensor for three hours, I set a toggle helper (input_boolean) to be true, and send a notification to my phone. If the toggle helper is true, my automation uses the alternate sensor instead of the main one. Once the value of the main sensor changes (eg I fix the communication problem), the toggle helper switches off, and the automation starts using the main sensor again.

Obviously, the “3 hour” setting could be adjusted depending on the typical fluctuation of the temperature you’re measuring.