The money spent on powering the NUC vs the PI is a speed and reliability luxury worthwhile in my case.
The design and implementation quality of HA is pretty amazing. In particular when considering the rate of change, improvements, and reflecting on the many third party variables and dependencies. From a reliability perspective, I’ll take almost any “settled” release of HA (settled means the fixes usually coming within a few days of a new release) against almost all other released software I encounter.
The main focus of HA is not industrial use, but home. I do require stability and reliability. For me, rebooting is fine when doing development (after getting PI replaced with a NUC/SSD the time it takes is insignificant). Who knows, maybe HA evolves into that; it is all interpreted…
One of the costs of rapid change is: it takes time to keep up. The chosen approach for how to handle updates (after the development phase for a site have passed) is important.
A benefit of rapid change is: we get new and relevant stuff in a timely manner.
There are always things that could be fixed or made better, but to me, the big picture of where HA is at (work done) and where it is going these days is more attractive than any of the Closed alternatives.
I have programmed for many, many years in a variety of machine and higher level languages. Been doing HA for almost one year, house automation for longer…
Thank you to all the HA creators and users contributing !
I don’t think HA has EVER crashed on me. (I run a Pi 3b)
I have had a couple of breaking change issues on updates, but who hasn’t ?
I think the longest uptime I’ve ever had is about 3 weeks (holiday, and I was the one who took it down) other than that it’s a restart because I’m changing this binary sensor or adding an input_boolean… You get the jist.
To be restarting/rebooting twice a day, of perceived necessity, is just crazy.
What do you have that makes your system so unstable ?
Maybe you should be looking at that instead of a new system ?
I started out with OpenHAB to use Homekit with it. I then switched to HA for a million of reasons (but mostly because of the development speed).
I use HA as my primary system now (OpenHAB days are long over and Homekit is something I have never used again. My interface is a 1000 times better than Homekit:joy:)
For what it’s worth I started and gave up on Home Assistant a few times before finally trying the Hass.io approach, which I’ve stuck with. It made setup and configuration just easy enough.
I know I said I’d respond with what my issues are quite some time ago, but I’ve been wrestling with how to express the problem when really, I think my biggest issue is that usb boot isn’t better supported.
My current system is running, but highly compromised due to what i’m 90% sure is traceable back to running a constantly updated database on an sd based OS.
As it stands… my install is jacked and needs rebuilt from scratch, but I’m trying to hold out for Pi4 usb-boot and hass.io usb-boot because I’ll be doing it again when those things happen. …and since the install is compromised, I’m not too interested in fixing the things that are broken or improving my UI, etc…
You can always boot from USB a Raspbian lite image and install hass.io on that as a generic linux install. I was running this over a year ago before I got my NUC so there are and have been options to do this for a long time.
Right, but then you get into software bloat for no real good reason… and having it run on a generic raspbian system is tempting me into adding services on to it that I’d prefer not to be adding.
I’m already not a huge fan of running through Docker, but at least effort has been put into minimizing the underlying linux host OS.
That said… limited options are limited. I’m trying to hold out for native usb-boot on a pi4 but even that is at a snails pace (at least they seem to be making pretty solid progress on network boot).
I’d give a VENV based setup a try, I’m using Raspbian on a RPi4, whole OS is on SSD which makes things far faster, however I still need a small SD card for the boot partition.
HA starts as a SystemD service, and config_check is much faster for me than it was running under Docker/Hassio/hassos.
As great as they are Pi’s do have a real limit on processing power… and my thoughts are that effort has been put into selecting an underlying linux host OS that doesn’t have the overhead of a full raspbian or even raspbian-lite, then docker is used as a compatibility layer for HomeAssistant across the various platforms. meaning effort has been put into making the underlying install as minimalist as practical…
To accomplish the same thing… it’s like… I can put together a device… raspberry pi + usb->mSATA addon card + 2.5" laptop HDD… and have something that functions like an AppleTV… roughly the same size… way more raw processing power than… but still doesn’t perform as well as.
By the same token… raspbian+homeassistant =/= hass.io
there are superfluous background processes and data stored… capabilities that will never be used in that application of a raspbian-based system…
I looked at a bunch of options, and it really came down to HAB or HA. I installed both on a Raspberry Pi 3B+, one after the other.
In the end it came down to a few things. First, I noted a LOT more activity on this HA forum than the comparable HAB forum. Looking deeper, there seemed to be fewer questions which went unanswered, and more in-depth answers here.
I’m still not sure I have enough experience to judge the underlying architecture or the differences in the logical approach taken by each, but the things I read about them made me think HA would be a better fit for me.
Today I’m struggling with some very odd behaviors with a few new automations. They were set up by doing a simple copy and paste from one that works fine. One of the three also works fine. The other two don’t. Yeah, it can be frustrating. But it’s forcing me to go back and learn some things at a level that didn’t sink in the first time around. I’ll probably come out of it smarter and with a more efficient system. I bought the Raspberry Pi just to learn, and hadn’t even decided to settle on a home automation project at first. So for this, HA meets my needs.
The headlong race to version 1.0 can be dizzying. I share the concern mentioned above that documentation is taking a back seat right now. The help I get here is fantastic, but a lot of it wouldn’t be needed if the documentation was up to date and clear enough that someone without a wealth of experience in HA and all it’s constituent parts could follow.
And, much as I love tinkering, I believe that HA will never attain it’s highest potential if users are forced to become experts in all the technical minutia under the hood. The user experience needs to mature quite a bit to match an accepted level of technology today. Or, as Taras said above, 20 years ago.
For perspective, I recently spent some time learning about Hubitat. There are several deficiencies in Home Assistant that, over the year I’ve spent time with it, I’ve learned to ignore (and accept the fact that unless I take the initiative to fix it, I can’t complain too much because it’s an open-source project).
After watching several Hubitat videos, I learned that some of those deficiencies have not been so easily ignored and have been addressed:
You don’t have to restart the application every time you add a new device or automation (proving Premise’s ability to do the same is not a ‘lost art’).
You create automations (rules) using something that I hope Automation Editor will eventually become. You don’t have to reload to make new, or modified, rules take immediate effect (another thing Premise achieved 20 years ago).
Common automations can be created by dedicated apps. For example these common activities can all be performed using purpose-built apps that guide you through the process:
scheduling a thermostat
activating lights based on motion detection
configuring a security system
In fairness to the Home Assistant project, Hubitat is a commercial, closed-source solution using proprietary hardware (so like SmartThings but with local processing, not cloud-based). There are inherent risks involved to users in the event the business ceases to exist. Not exactly an apples-to-apples comparison but, nevertheless, an example of how competing home automation solutions have addressed certain issues.
I briefly dabbled in OpenHAB before coming here and this point very nearly turned me back to it. It feels so 1980’s to have to restart the entire system after a small change.
I’m glad I hung around though, I can’t remember a more helpful and generous on-line community. I wonder if that will continue as HA grows?
PS It must be feasible to get over this because we do have the ‘reload’ options. It’s just that I have had some very bad experiences using them since they became an (unannounced, why?) option when using packages.
It only became available to users with packages a few version ago. I don’t know when because it wasn’t announced and I discovered by accident when someone mentioned it in passing on here.
I made a feature request well over a year ago - it doesn’t exist anymore though presumably because it got deleted when the feature was (silently) implemented.
Klogg, I did not doubt your word just the context.
I must say your previous thread was an interesting read, bookmarking a significant step change.
I admit that my first dabble with HA was prior to such and I’ve only just come back following a major house move and building project.
I must thank people like you for the efforts required to knock the rough edges off, here’s to a more ‘spherical’ future
Cheers
If you monitor openHAB’s log you’ll see that at the moment you save changes to (for example) the Items file, the log will show that the modified file is reloaded and processed. In other words, they’ve implemented a ‘file watcher’ that automatically reloads (and validates) modified configuration files. Although the entire application does not need to restart, it does reload the entire Items file.
In contrast, Premise allows for incremental changes. I can add an entity, or modify an existing one, and only that specific change is registered (other existing entities aren’t reloaded). It also applies to automations because it uses its own text-editor with code-validation. If you make a syntax error and try to save the file, it will complain and highlight the error.
I don’t own a Hubitat, so I can’t confirm it, but based on videos I’ve seen, it appears it also supports incremental additions/modifications of devices and rules.
I’m familiar (to various degrees) with about a half-dozen home automation programs. In addition to Home Assistant, the other one that also needed restarting was Misterhouse.
Ha! Thank you!
But I am under no illusion that I had anything to do with the improvement.
I am not at all sure that anyone who matters actually looks at the feature requests on this forum very often. As I understand it all the ‘important’ noise is on Discord.
since your Premise examples make our mouths water, I wonder did you ever directly talk to Balloob about this? I mean, posting on this community hoping it will ‘enlighten’ the dev’s to adopt any of the many great features would be idle hope I fear.
I would imagine this to be a desirable outlook for Homeassistant and would be keen to know if this is felt the same way on the dev’s side.
It depends, and you may sometimes feel that your voice is but a gnat cheering at a football stadium, but when you represent a groundswell of opinion, or the soundness of your opinion is actually heading in what the devs feel is the direction they should be going then perhaps you made it happen just a smidgen faster.
If your voice is heard by ‘some’ individuals (eg Tinkerer, Taras, pnbruckner, Petro, finity, Tom. and the like (many, many others) (they are ALL OVER this forum and have helped many many people) the justice and logic of a good request MUST make it through sometimes.
Taras is always ‘considered’ and clearly ‘well intentioned’ .
This thread would be heresy on some ‘software sites’ where the colour of your pajamas must meet certain criteria for you to be able to post (I think this was mentioned earlier in the thread) So I’m confident that ‘someone’ is watching. The question is if they are laughing or saying “well, if it has been done, why couldn’t we ?”
I get a bit frustrated when I see that an integration has been added for a new component that will hoist a flag if its “Tuesday on Pluto and Thursday Venus but only if Capricorn is ascendant over MacDonalds” - leaving Teras’s suggestion gathering dust.
So I for one will help, where I can and cast my vote wherever and shout outside the gates of Mordor, (as you have) to help bring this (and others) to bear fruit.
Alternatively we just assassinate all the devs who more interested in playing with their navel
No, I know that each of these changes is huge and the devs don’t have limitless time and can’t just wave a wand.
But shouting encouragement is a mandatory component for change - where do I sign up ?
)
and I discovered by accident when someone mentioned it in passing on here.