What’s your fallback protection?

So far, I’ve mostly just used HA to augment the existing kit in my home. If my HA box dies, the Hue hub will keep turning the lights on and off, the Evohome controller will manage the heating, etc.

But, of course, HA is becoming ever more useful, and I often think about what would happen if the disk ran out of space, or the SSD or power supply died, when I was on the other side of the world and my (non-techie) wife was at home…

Do others worry about this too? Keep a spare Pi with a duplicate SD card in the cupboard so you can hot-swap it?

I’m now running on a NUC, with the database on my Synology, which should make each bit more reliable than my early experiments on the Pi, but it means there are more things to go wrong and they’re harder to replace.

How robust is your home automation in case of failure of a key component?

2 Likes

I run HA on a NUC in Docker and everything I need to run HA I can move to a RPi that I have sitting on the desk running an equally updated HA version. All I would have to do is transfer my config directory and my zwave & Zigbee sticks over and I should be up and running again.

And if all else fails I can still run everything the old fashioned way with light switches and such. :wink:

I’ve been tinkering with home automation systems for a couple of decades now. My rule of thumb is never, EVER get rid of the manual control.

2 Likes

:slight_smile: That’s probably wise.

Though I’ve been running for quite a few years now without any light switches — at least, without any that didn’t go via Zigbee and the Hue hub.

But my Hassio setup is rather more complex. The question is whether I could tell my wife how to get it going again if I were on the other end of a transatlantic phone line.

If something more serious happened to me, she’d probably just have to move house :slight_smile: .

Perhaps I could get several Raspberry Pis and run HA on a Docker Swarm…

If my mini PC (Hassio NUC image) was to die I guess I could run up a VM on my office PC. Might be worth trying this before I go away for a year.

But yeah all my stuff is local control so no drama with lights or heating if this does not work.

Mine did die.

Long story short. Make it easy to use manually. Automation should be for convenience but users should be able to manually override easily.

Ah, thanks, @silvrr - that’s a good thread too.

I realise that I’ve been concentrating on backing up the YAML files, but that the entity and device registries are just as important: it would be quite a pain recreating those. So it’s a pity they’re JSON files tucked away in a hidden .storage directory.

Discovery is very useful, but it would be good, I think, if, once discovered, one could freeze certain bits of configuration in human-readable YAML files and the system would load those in preference to the .storage files. Is there a way to do that? To say, for example, “this zigbee device discovered by deconz is actually binary_sensor.garage_door” ?

I have had to edit those JSON files in .storage in the past, but I have a feeling that’s not what we’re meant to do. :slight_smile: And stuff that’s important and persistent probably shouldn’t be stored in a directory beginning with a ‘.’.

We do have a file called known_devices.yaml… but I think that’s just used by device_tracker?

Fallback is manual (get up and do it myself)

I considered having nginx failover to second server in case HA went down but this would require HA and nginx be separate boxes and not likely server will go down. I concentrate more on HDD/server backup plan. Data loss is more important to me than short term server down.

If server uptime becomes concern I think cloud strategy or mix would become best. Cloud as in move my server to cloud not Nabu Casa cloud but maybe that could be the way if it was concern which I doubt it will be.

I thought through fail scenario for when I’m on the road and always comes down to “can I login to a server via ssh somehow or remote”

If you have remote capability to secondary machine I think all is Ok except complete power or WAN outages

Snapshots in Hassio back these up too.

I hope so.

Ahhh! And they go into the backup directory, which is conveniently exposed by the SMB server. This makes it trivial to add them to the Active Backup configuration on my Synology - thanks!

1 Like

Is that software running on your NAS?

Every time I learn more about Synology I hate my uniformed choice of a WD EX4100 a little more.

Yes - I’m becoming something of a Synology fanboy now :slight_smile: That’s the new Active Backup for Business software (which, despite its name, is free), and works very well. There are other older backup packages too which provide different facilities, but this is very good for backing up from SMB or rsync.

My home NAS is a 5-bay one, which seemed extravagant at the time, but I was buying a new iMac and I asked myself, “What’s more important, this machine, or the data that’s on it?” And that’s when I realised I should arguably be spending rather more on my backup solution!

I have a second, off-site, cheaper (2-disk) Synology, which backs up my backups :slight_smile:

But it runs the same software - I do recommend them, even if you just get one of the cheaper ones. It’s never missed a beat.

I also have mine plugged into a UPS, which has the interesting side benefit that I can monitor my mains voltage in HA :slight_smile:

54

This has happen to me: twice.

just a couple of days before returning home after a long vacation, my HA died and there was no way to turn on the

heater and had to come home to a cold house. WAF went straight down to null.

So for this years vacation I’m planning to do:

  • Move DB to Synology (done)
  • Move logs to Synology (log center)
  • Move HA from PI to NUC (in progress)
  • Use the PI as fallback with the following features.
    – 10,000 mAh power bank.
    – add a secondary connection to internet (neighbors wifi or LTE modem or both)
    – Install a fresh copy of HA on docker and configure the must basic stuff only (lights and heating) no automations or any other docker
    – add the ability to monitor all the other devices (NUC, Router, Internet)
    – add the ability to reboot HA on the NUC
    – add the ability to reboot the NUC via SSH
    – add the ability to reboot router via SSH
    – add the ability to power cycle all the network infrastructure via wired smart plug (relay)
    – ultimately, be able to control at least the heaters when returning to home.

The plan is to keep this PI as dumb as possible, so any action must be initiated by me, at least at the beginning.

2 Likes

Ah, putting the logs on the Synology NAS as well is a good idea; I might try that.

I back up my HA user folder, along with other critical config files for things like LIRC, to my Synology NAS.

For remote access I use duckdns, and as a fallback point I have a VPN I can connect to that puts me on my home network. A 3G modem acts as a secondary network connection should my copper connection go down.

In terms of hardware, I have a primary and a secondary that I can copy the config to and be up and running again in minutes in the event of hardware failure. Power-wise I have nothing, but looking to add a UPS that can support my network equipment and Home Assistant.

Primary:
UP Squared board (Intel Atom N4200, 8GB RAM, 128GB eMMC)
USB UIRT

Secondary:
UP Squared board (Intel Atom N4200, 4GB RAM, 32GB eMMC)
USB UIRT

I run HA on my home server (10+ year old gaming PC) in it’s own dedicated VM. All of my VMs and CTs are backed up weekly and saved in cloud-synced folder (Yandex in my case). Right now I don’t have any hardware to fallback to, but if it failed I"d be able to get new server and restore VMs from snapshots.

If I really needed to have a fallback I would have a second Proxmox server and it’s possible to setup replication and failover with Proxmox tools I belive.

But as many people already said it, I just try not to rely 100% on automations, I always approach it in a way that it should be possible to control everything manually. And simple ZWave and ZigBee associations might still work as they are stored directly on the nodes themselves, at least from what I’ve read recently.

In about 3 years I only had one time that server failed, but I was at fault myself, did not run the system update properly and had to reinstall everything. At the time I was running Ubuntu and now I’m actually glad that it happened and forced me to finally try out Proxmox and I fell in love with it :smiley: That time we spent about 3 days without Hass… It was torture, we had to get off the bed to turn those lights off :astonished:

does someone have a config like this:

HA Server (master) on a RPi
HA Fallback Server (slave) on a RPi

When edit configuration on the master one you can give command to copy everything to the slave server.
Every 3 or 6 hours database will be copied to slave as well.

When master crashed, easily change (in my case) msata ssd to the master. Startup and you’re running again.

Reconfigure the crashed one and put this one back in the slave. Update master with the scripts to update configuration and database file to slave.

slave become master then. Old master become slave

Theory looks nice my opinion :smiley: … only im no developer so can’t write the scripts haha