I see all of this discussion about snapshots and SD cards failing and undocumented backup requirements…
When I was running HA on a Pi several months ago I kept a manual backup of several versions of my config directory on my PC and before and after every major update or every periodic update of HA I would make a complete SD card image and save three versions on my PC. That PC had automatic backups to the cloud using crashplan.
I never had to worry at all about screwing up my config so bad I couldn’t recover it fairly easily. And if I screwed up the Pi OS badly enough or the SD card died I just popped in a new one and re-burned the image and I was back up and running in less than 15 minutes.
And now that I have moved to using Docker (on a NUC but that point is irrelevant) using Syncthing I get completely automatic backups to my PC which auto backs up to the cloud with versioning. And you could still have the ability to backup your entire SD card to your PC regularly.
And I still have the ability to add any other software to the Pi or Docker as I see fit without being reliant on someone creating an add-on for it.
You can’t do ANY of that using Hassio. You are completely locked in to that OS and only HA and the add-ons.
I really have a hard time seeing the benefit of using Hassio. Especially when you realize that, even tho it’s supposed to be plug and play, you still have to configure the add-ons to work with your system. That’s not plug n play any more than Docker.
High Endurance SD cards - is that a real thing? By which I mean is there any kind of standard so that one can verify that it is indeed High Endurance?
I ask because on my first search on Amazon I found a SanDisk High Endurance 32GB for £12.35 and a Transcend High Endurance 32 GB for £22.04. Are they both going to be the same spec?
I’m an old-dirt unix guy from back in the late 70’s. I kinda like the hassio thing, no need to demand those that want to join in to understand the intricacies of Unix before they ask questions.
With that said, I highly recommend to the new-dirts to at least figure out how to set up Samba so they can, at their option, manually copy their config folder to their pc so in case of total meltdown, they can resume with where they left off.
And everything is pretty much no different operationally-wise using hassio or docker. You still have to set up any additional components to use with either install method.
Until something goes wrong… At least with Linux you get the benefit of having 50 years + of wisdom to draw on to fix things. With HassOS you have…this forum. And if Hassio dies (not the card but just the HA part) then you’re locked out of your system until you reburn a card since Samba and SSH rely on HA starting up to function. And hope the snapshots are OK to restore from.
A huge amount of ignorance mixed with a few facts in your post, you might well be whizz with Linux, and that’s great, but you don’t know much about hassio.
Rather than just ‘seeing’ discussions you might want to actually read a few of them. Power users use hassio in Docker and therefore have the potential to do everything you say they can’t but equally users who have no idea with linux can create quite complex systems without frying their brain.
I’m currently running hassio in a linux VM on a Windows server and backup both snapshots and the whole VM which you’ll probably look down on but I’m happy.
Best source for comparisons is dashcam communities as that is one of the worst environments for SD cards.
The Sandisk HE is generally the most recommended HE card as it’s often on offer around the price you found. Ironically, ‘normal’ Sandisk cards aren’t recommended at all for dashcams either because they just don’t work or have reliability issues.
I’m no linux whiz at all. I never really used Linux at all aside from dabbling in using a RPi for Retropie. pretty much everything I’ve learned in Linux I’ve learned in the last year using HA. And I’m STILL FAR from being competent in Linux.
I think your post kind of makes my point, tho.
I’m glad the Hassio Whiz’s can figure out ways to work around the limitations inherent in Hassio (if you’re modifying Hassio to work in ways other than originally designed then you can’t really say you are running “hassio”). And it took some other Hassio whiz to come up with a way to run Hassio in Docker, too.
The whole point in creating Hassio was supposedly to allow users who weren’t “whiz’s” at any of this to easily get up and running. But it’s not really true to imply that installing HA via Hassio will be all plug 'n play, unicorns and rainbows. I think there are more posts on here dealing with issues with Hassio than other install methods. Maybe because it’s installed by more people who have gotten the impression that it’s the “standard” way to install it (i’ve been in at least two threads trying to explain this exact thing in the last day or so). But then when things don’t work right they don’t have the tools to try to troubleshoot it themselves.
I have no issue with anyone who wants to run anything however it works for them. I honestly, really don’t. If it works for you then more power to you. I’m glad you’re happy!
No it didn’t. Hass.io ONLY runs in docker.
Installing hass.io on a generic linux server is supported and documented! It’s no harder than any other HA instalation.
I am running via this method. I also used it on a Raspberry Pi using Raspbian - full hass.io.
I LIKE hass.io. I like that I can use the provided add-ons and they are easy to configure and well documented and supported and upgraded via hass.io which is one of the things many people like about hass.io
However I also run a bunch of other containers with other software applications and it all coexists and works EASILY.
So please don’t make assumptions about hass.io and look down your nose at people that like it!
I have the best of all possibilities the way I am running.
You can find industrial grade SLC based SD cards but most people shy at the price. Last I looked about $160 for a 32GB card. But you get something good for >200,000 write cycles versus ~4,000 for the consumer grade SSD cards. I know of industrial machines running for years with such cards and write heavy environments without problems.
???
The hassio image is available via update button in the frontend for all hassio a day or so later than other installs.
Spam the dev boards? I have no idea what you mean.
Beginner’s question here, prompted by the comment that sometimes a restore from a snapshot is subject to failure. Not questioning the fact, just asking if that were to happen, could someone get a new hassio install from scratch, reinstall the addons (I only have 5 so that can be done quickly) and copy/paste the \config folder (perhaps with the exception of the dB file) using samba back into the PI. Would that work or are there other pieces missing that would make the new config fail ?
You wouldn’t copy the database. Yes just copy back the yaml files and that will work but of course you need to reconfigure the addons manually in that case.
It sounds like you have found that with great power comes great responsibility. That being said, I would not be running the grafana and DB apps on an SD card. I have a suspicion that they contribute to SD card failures due to the way they function. That being said, there is an automated way to do backups. Rob from “thehookup” on Youtube had a video about it.
to see when the update is available and then do an update via ssh. I can usually do that a day before the frontend reports a release being available and prompting for an update.
In the case of 0.81.1, the docker tag is incorrect and when I updated (and many others as well on hassio) it’s saying 0.81.0dev0 so yeah… that prompts a few discord questions.
True but it is very vague about that fact on the HA main website.
In fact, it’s nowhere to be found on the “getting started” page. From that page you wouldn’t even know anything about non-hassio installs without digging deeper and unless you are an “advanced user”. No wonder there is confusion among new people when others talk about other non-hassio situations.
And that install method is buried pretty deep on the main HA website, links to a procedure on github and with warnings like the following:
I’m sure a new person who isn’t very well versed in things like this and just getting their feet wet in DIY home automation is thinking “Sure, that sounds exactly like the thing I want to try!”
It would be interesting to know when that github procedure was added. Was it there since the introduction of hassio or added later?
Let’s try not to straw-man me on this please.
You can like whatever you want. I really don’t care. I really don’t. No, really…I really, really, really don’t.
I don’t think any less of any person on here or anywhere else in the entire world that uses and likes Hassio. I really, really, really don’t.
Man, people sure get quite a bit over-sensitive when you step outside the group-think, poke the sacred cow and point out some potential issues that new people are GOING TO FACE.
I completely agree that the explanations about different install methods and platforms are very confusing - particularly the differences between venv’s, hassio etc are concerned and it would be daunting for a new user. It’s not at all spelled out and there’s hundreds of posts asking where’s my hassio menu when they installed hassbian and they just don’t understand the difference because it’s not really explained very well. Adding to the confusion, you can install home assistant in docker as well or you can install hassio in docker and those are different experiences.
The point of Hassio IS to allow non-whiz’s to easily make it work and non-whiz’s will generally be happy to start with a RPi and that install is pretty simple unless you want to use WiFi. I think the problems you allude to are then people using hassio who suddenly want to install other linux stuff because of course on a $25 device you should be able to run resource hungry things like Hassio and then also torrents and vlc and kodi as well right… That is where people run into trouble - not understanding what each install is and it’s limitations. I guess start off basic and then build on that knowledge as you move forwards.
I know some people have attempted to improve the docs as they have been through these frustrations but they just tend to get shut down.
Some people do consider people using hassio to be inferior on here. To me it’s the best of both environments the way I am running it and it’s not that complicated - I did start on HA with an all-in-one install just over a year ago. When that was depreciated, I decided to try Hassio. After that I decided to try raspbian and install hassio in docker on that and have since moved to a nuc. The good thing is you can move your configuration between all of them pretty simply. I could never have started where I have ended up when I first started so as your knowledge grows you can easily change…
As some of you advised, I dug in to the snapshot and look for config files and it worked! There was complete configuration.yaml in the snapshot, just copy -> paste and voila. BTW, I still have no idea why complete snapshot restoration failed at this point. Another strange thing is, that in the most recent snapshot have fully functional configuration.yaml in it, but no sign of ui-lovelace.yaml but snapshot which i made a week earlier contains both config (which worked only partially) and ui-lovelace.yaml. Anyway, my main config came back to life, same as all of my automations also entities history started to work again. Still have some issues with InfluxDB and Grafana, probably I need to focus a bit more on that.
