Because it’s easier to put mariadb on an existing utility vm than moving everything to homeassistant or struggling to get hassio to run in a debian vm.
I don’t accept that one is easier than the other. Running hassio on linux is very easy IMHO.
Seriously? Don’t accept all you like, but I don’t like to waste my time. 1 minute to look up how to apt-get mariadb followed by 5 minutes of config and done vs setting up my homeassistant all over. There’s enough fooling around with HA I’d rather avoid yet another full on config session.
To be fair, installing HassIO in linux isnt any more than 1 minute either, it can be as simple as running just 6 commands.
sudo -i add-apt-repository universe apt-get update apt-get install -y apparmor-utils apt-transport-https avahi-daemon ca-certificates curl dbus jq network-manager socat software-properties-common curl -sSL https://get.docker.com | sh curl -sL "https://raw.githubusercontent.com/home-assistant/hassio-build/master/install/hassio_install" | bash -s
The config is 100% transferrable from installation to installation. You don’t have to reconfigure home assistant
I had issues with Hassio for a while on an RPi3, every update i would have to reinstall to the SD card. Just wouldn’t reboot for whatever reason, anyhow, I switched my Home Assistant using docker-compose on Ubuntu Server on an old pc collecting dust and haven’t looked back. And I knew next to 0 about setting up Home Assistant outside of hassio, and even less about linux, try to find some guides, it really helps. Easier than I thought it would be.
I’ll have to check out using mariadb for this. Never occurred to me to use it for the HA database, good idea.
Sure you can install it, but that doesn’t mean you’re up and running. It’s never as easy as that. For example, now you have to fight with getting your USB z-wave/zigbee sticks to talk to the VM properly or setup some alternate blue tooth hardware for presence detection, for example.
It’s much simpler just to offload the db to a machine where constant writes won’t do any damage.
It’s easy to do and I found it significantly improved my boot times as the homeassistant.db starts growing. I really should figure out why mine gets big, but out of the box with logging turned on it grows too fast.
Agree. And reboots are SOOOOO much faster.
But I ran Hassio on a Pi for more than a year. The only problem I had was when my history db filled up my SD card because I was retaining too much historical data. Moved that to MariaDB and it was fine on Pi for a long while after.
I find it funny people trying to give the OP advice. He was trolling us, because if he had actually cared to fix his issue, he would have listened to people in his other thread, instead of arguing with the solutions provided.
He doesn’t care, he wanted to make a grand exit, and troll us. After seeing the way he argued with everyone in his other thread, I don’t care if he is a part of this community or not.
No issues here. Running Home Assistant and Ubiquiti Controller on the same RPi.
> $ uptime > 17:38:18 up 279 days, 4:35, 3 users, load average: 0.01, 0.02, 0.00
I read through the thread and I want to say to all the people that replied in a hard tone that we are only humans. Both the developers and the newbie users. Developers are not perfect we get more than we pay for, yes, for sure, all correct. But the poor sods that try Home Assistant and see one SD card die after the other are only humans and we should all allow some human steam to escape. Just ignore the mood. People need to steam out sometimes. They will be good again and thankful to the community once everything works.
You all gave a lot of good advice. Very good advice.
This and other similar threads must raise a question to lead developers to reflect on an important decision point and ask themselves this question…
Is it wise to push very hard the newbies to install hass.io on a Raspberry Pi using the recommended installation? Most of the answers in this thread suggests that it is not wise. So why it is the default?
Could Hass.io run well on a Raspberry Pi with SD Card if the default out of box settings were less aggresive with the read and write?
I have played with Raspberry Pi’s for years but not with Home Assistant. I have seen SD cards die often. I have learned my lessons. For the past years all Raspi projects I do are with a small cheap external USB SSD. You get a 64 or 128 GB for nothing these days. It is not fast via USB2 but faster then the SD card and for sure more reliable. (Except Western Digital MyPassport USB SSDs. I have had trouble with two. I think there is some driver issue). But anything else I have tried worked great. The latest Raspis are easy to use with USB driver. The previous model had to have a little hacke-di-hack to enable the feature but they can too.
And here comes the problem. You can download any Raspberry Pi image I have ever tried and program it directly to the SSD using Etcher and it just works. Except Hass.io. It will not boot from an external device when you use the image. The developers must have hardcoded some device names because there is no reason why it should not work.
You can install Hass.io using the manual method where you install Raspian Lite first. And then follow the installation instruction for how to install Hass.io on Debian/Ubuntu. Raspian is Debian based. The installation follows the instruction 100% and it works great. It may look complicated but just follow the steps and you are running in 10 minutes.
I 100% recommend this method instead of the recommended SD card version. But you cannot blame the newbies from followed the steps they read on https://www.home-assistant.io/getting-started/
It is spelled out to follow the recommended installation - on an SD card. But it is a not wise unless you make a hard effort in learning how to stop logging and history. It is enabled by default.
I really encourage the lead devs to rethink the beginner strategy. A combination of a default setup that load the SD card less and an image that can install on an external USB drive with a recommendation to do so would be a low hanging fruit to a better newbie experience.
I personally ran my Hass.io on a Raspi for 2 weeks and then I bought a NUC (the cheap Celeron based one) and it works so smooth and great. But it is good to start on a Raspberry Pi to find out if HA is something for you. And if your setup will remain small and simple, running for years with a Raspi and an external cheap SSD is for sure possible and OK.
That was my 2 cent. Thanks all for the great project. I am two weeks old - just moved over from Homey with a large home automation installation. I hope to eventually contribute with what I can with the skills I have.
I think the vast majority of issues on the default/recommended install issues would be solved by simply turning logging/recording into selecting what you want to record rather than the current default of log/record everything.
Such a mature and wise comment, Kenneth. Completely agree with you on the point of allowing people to blow off steam, which they inevitable have when running into one failure after another. I’ve been on the verge of getting really angry myself with annoying behaviour of HA that I’ve observed over the past couple of weeks.
I too agree that there should be no recommendation to use PI + SD or Hass.IO should support external storage for boot / data / whatever. Unfortunately I myself didn’t spend too much time looking into details of HA - just read various reviews / tutorials, which never said about any issues with PI. I trusted reviews and didn’t have time to look into small print, which is not easy to discover at first. So I’ve purchased a PI specifically for HA. Now it turns out that PI is not good enough and I need a NUC or something like that. And that is a disappointment for me because running home automation of a regular computer just doesn’t make sense - i want something small that I can leave powered up when I’m away for a week or a month / two.
To all those with comments about “it’s a free software so shut up” - even free software should be designed / distributed in a way that benefits people and doesn’t result in major dissatisfaction. It’s always about your customer base, no matter whether you charge money or not. And it’s only logical to address the foundation before running off into the wild with features.
I agree 100%.
The thing is, this is an open source project.
Everyone here, including yourself is free to make changes to the public documents, as well as the default configuration to make this a reality. Home-Assistant isnt a closed box where only special people can make changes, or we need to wait on some nebulous group of ‘others’ to implement a change.
There are only “lead developers” in the sense that balloob (Paul) is the only one getting officially paid to work on this. Everyone else are just volunteers.
Every time there is a comment saying that “The docs should say this, or say that” or “The default config would be better if it was XYZ” could also be time spent implementing those changes.
If you’re not sure that going right into changing these is the best path, you could open it up for discussion on the HA architecture discussion github page.
“Be the change you wish to see in the world” is more than just an empty phrase sometimes.
He had a previous thread regarding these issues where people tried to help him. He did nothing but argue and didn’t follow the solution suggestions. Then he opened this thread and gave different information about how long an install has lasted and how many sdcards he’s blown through.
He’s just looking for attention and we gave it to him. We fed the troll.
So change the documentation?
Have you thought about bringing these ideas to the devs via GitHub issues?
Very wise words, however it isn’t the OP’s first thread about this nor were the people very hostile at first. But still I agree with you. However, I might have to mention the following, this project isn’t a beginner project and is pretty well documented (I am a beginner ). The reason I say this is because in my opinion to make the most out of Home Assistant is to have basic knowledge about coding but even more importantly so is to have basic knowledge of hardware in general. Most people here will know that SD cards are either slow or unreliable. You’d expect one to have this knowledge already when starting a complex project like HA (no offense intended). It is like buying a graphics card without knowing if it even fits in the case (if you know what I mean). I understand that the OP uses a raspberry, and even though I have zero experience with those devices I know that SD cards have a limited lifespan (just like hdd’s and ssd’s do, but way shorter).
I also addressed the OP to send me a PM, because the OP is Dutch (like I am) and I was (and still am) willing to help him/her in his/her native language which would make the process probably a lot easier for him/her. And honestly I hope he still does, because no one should be able to miss out on this great piece of software (I have tried OpenHAB in the past, but HA is much better imho).
Anyways, long story short, in my personal opinion I don’t think this project is for beginners (even though I am one) at least not if you are not willing to devote a lot of time in it to understand how this works. There are easier solutions like for example Xiaomi Aqara (which works perfectly fine with their proprietary app and is easy to setup) or sonoffs which are easy to setup as well and their app works great. Only downsides are that you don’t have it locally and that you can’t use a single app to control all the devices (however Google Assistant, IFTTT and Alexa works on both of these products out of the box, and Xiaomi even supports Homekit and Ikea Tradfri).
But I agree that people shouldn’t flame beginners for whatever reason or question. I always say, there are no stupid questions, only stupid answers.
I’m running HA in a raspberry more then a year without a single problem…
Probably is something that you are not doing right… don’t blame an open source…