Why I'm leaving for home assistant


#1

I just wanted to leave my thoughts here, with the hope that (a) some people might benefit from understanding the strengths and limitations, and (b) the Hubitat team might get some insight into either marketing or feature work, since I initially thought this hub was perfect for me, but here I am a few months later.

There were quite a few things, each of which would not be a big deal on it's own, but in aggregate ended up pushing me off the platform.

First, Hubitat is a great platform. I have no doubt that it's comparable to (if not better than) SmartThings, and everything I write here probably applies to any commercially available HA solution. I had a Wink Hub before this, and it was miles behind in capabilities.

Second, Hubitat has impressive hardware support. Almost all my devices paired with it almost immediately, and I had no issues with disconnections or delays.

But compared to rolling a customised solution on Home Assistant, there were a few shortcomings:

  • SSL was added very recently, and it doesn't support automated LetsEncrypt renewals. So I would have to set up a second machine that would script updating the certs (assuming custom certs show up soon). And even with SSL, there's no support for advanced authentication (like totp, or client certificates), so I ended up running it behind a reverse proxy.

  • No built-in WiFi, which meant I had to provision a dedicated wifi-to-ethernet bridge machine (same one that runs the reverse proxy, and handles SSL certificates as well). But that adds more complexity for network devices (e.g. Wemo, Google Cast), because of double NAT. Home Assistant runs on a RPi, which not only gets me built-in Wifi, but also stronger firewalls, eliminating the second machine for the reverse proxy while simplifying SSDP and mDNS.

  • Rule editing is a very manual process. For example, I have two bedside remotes, and I wanted each remote to have the same button mapping. Couldn't find an easy way to duplicate the settings. In the end I settled for a second remote simulating button presses on the first remote, and saving all the logic once.

  • Multiple actions for a rule are pretty painful to manage. In comparison, not only does Home Assistant allow YAML files for config (so I can generate them dynamically), it also allows multiple actions with conditional exits.

  • Backup and Restore is misleading at best. I've posted about that earlier, but in summary: losing the hub means setting up all the automations from scratch. In contrast, Home Assistant uses customisable Entity IDs for all automations, so you just need to rename the devices when they're first added, and the rules will start working.

  • Hubitat's web UI is noticeably slower than Home Assistant, even though they're running on comparable hardware (I think. I'm using an RPi 3B+). But even if they're not, I can always just upgrade the Home Assistant hardware, and move over the USB ZWave+Zigbee sticks and restore a backup, whereas no hub migration strategy exists for Hubitat, even if a new, faster version were to arrive.

  • The "magic logs" that only support can see, and the persistent (or backdoor) connection that ability implies makes me slightly uncomfortable, especially since Hubitat was billed as a local hub. I understand the necessity of a cloud connection for the dashboards and Google Home/Alexa integration, but turning both of those off does not prevent support from reading logs that even I cannot access.

Beyond these, I was also looking into making the Hubitat data more easily accessible for storage and graphing. I use prometheus(+influx) and grafana for other metrics, so integrating the devices in my home would be nice. To that end I started writing something that would scrape the master dashboard json and convert it to prometheus metrics, but Home Assistant includes prometheus built-in.

Home Assistant also supports many, many more devices out of the box, and IMO is easier to add custom components to. (Although it does require more restarting...)

  • Amcrest IP cameras work for motion detection, (and even show snapshots on the dashboard)
  • Lutron's Connected Bulb Remotes work with the deconz integration, whereas I just could not get them to work in Hubitat.
  • Geolocation can use my existing Maps Location Sharing system, without requiring a new background app on my phone. Ping based presence is also nice
  • Nest still works in HA, although Hubitat might decide to bring it back later.
  • Chromecast works incredibly well, status is always correct, and I can set up playlists to play on devices using automations. (But Hubitat is rapidly catching up here, their Chromecast integration is very impressive even today. One of my favourite automations was to turn off all music when the bedroom lights go off at night).

And this is totally minor and almost irrelevant, but I never really grokked the 'modes' concept in Hubitat. It seemed like I would always want a combination of modes, at which point I considered just using virtual switches with conditions everywhere. This could be because I wasn't using them correctly, but I wanted something like "day"/"evening"/"night" combined with "home"/"away" combined with "guests"/"no guests", so that I could write a rule that goes something like "away and no guests -> turn on recording for indoor cameras". That exact rule is trivial to set up in Home Assistant.

The Open Source aspect, as well as being able to version control the configuration definitely appeals to me, but that's not for everyone, so I don't see the lack of that as being a downside to Hubitat. But being unable to edit the configuration in anything other than the UI, which is often slow and requires many clicks, certainly put a damper on my enthusiasm.

To be fair, of course Home Assistant is not perfect. Zigbee lights took many attempts and many were unavailable for hours until I reset and rejoined them. The UI is still a huge WIP when it comes to editing automations or scripts, and renaming z-wave devices is painful by hand (but I have a script for that now).

Many devices are displayed in an unexpected manner, e.g. the Nest protects do not show a battery sensor, but a text sensor for battery that says "Ok" or "Replace", and requires some work to be seen as an actual battery sensor.

And setting up a new RPi with home assistant is hours of work compared to plugging in a new Hubitat. Customising the dashboard is a lot more work, too, and YAML can take some getting used to.

It also needs frequent restarts to pick up some classes of configuration changes, which can take a while if z-wave is in use.

But in conclusion, I think Hubitat is a great SmartThings alternative, but it's definitely not for a programmer like me. A lot of the reviews I read made it sound like Hubitat was a very do-it-yourself kind of hub (pointing to the user apps and drivers pages), but I incorrectly assumed that meant the Linux kind of DIY, which is not the case here.

I'm sure there's a lot here that most people don't care about, but if I had read this exact post before getting the Hubitat, I could have saved myself a few weeks, so I'm putting it out there for others to read.


#2

You make some good points. I particularly agree with you on the rule machine issue, setting up rules is beyond tedious with no way to create common functions and duplicate actions. I already have over 145 individual rules set up and I will be extremely upset if something happens to them.

Personally, I hope to use Home Assistant in conjunction with hubitat. The zigbee and zwave support in HA is nearly non-existent, with no easy way of adding support for new devices. So if you have any significant amount of those types of devices you are not going to like HA for very long.


#3

Hope you enjoy that platform. It does sound cool - I looked at it before going with HE. Decided I just didn't have the time and patience to deal with that kind of setup anymore but really like the flexibility and that you can use your own hardware..

My goal is automation/reliability first - I still have the WAF to consider.. and she is even LESS patient than I am. Wish I had more time with the million other things I'm involved with..

My HE setup is fairly simple - HE is on my "home" subnet and my firewall handles my VPN.

Also not sure why you'd want to expose the HE directly to the internet. I was under the impression due to the immaturity of the platform and state of security controls and my own experience that this was/is not a good idea. My VPN takes care of dashboard and UI access.. not that I really use them remotely anyway. However the hub is STILL open on my subnet and that is not ideal.

As a programmer myself I too find RM a bit cumbersome but workable. Just a bit different mindset. I have also been playing around with coding some devices and apps. There is potential there to do a lot of things.

Anyway those are my thoughts.. happy adventures with your new setup!!!


#4

I think the OP does a decent job breaking down why they want to switch and sorta shows pros and cons on both sides of the fence. I can see how at least some of the points are valid from what they want.

I think HASS still is far from mainstream usability. I recently watched a video for just installing the latest release of HASS, it took the guy about 10 mins, and everything he did looked liked it should be something the install does automatically.

While I love tinkering and learning, certain things I just want to work. IMHO, Home Automation shouldn’t be any more complicated than it has to be. There is a reason I don’t run all my systems on Funtoo or Gentoo; sure they could be faster, more efficient, more stable, more secure, etc, etc. But ain’t nobody got time to mess with flags and to compile everything. HASS still feels like way to much work.

I also am concerned a bit now that Ubiquiti owns HASS. While some may think that is a good thing, I have my reservations because they overall suck at maintaining and evolving products after release. In their latest release of the Unifi Controller they focused on a stupid “Dark Mode” UI update before basic capabilities that users have been screaming for for years and years. That may upset some, but it doesn’t take long using some of their stuff to find out some of it is great and the rest is half baked, half finished, lacking garbage that costs too much. I guess we’ll see how it evolves moving forward.

Ummm.... 145 Individual rules?


#5

First, thank you for a very nice write-up of the reasons for your decision. I can totally respect your points and I wish you well with Home Assistant.

I am curious... As a programmer, did you ever experiment with writing your own Apps and Drivers on Hubitat Elevation? I agree that RM can be a bit clunky at times (but it is always evolving and getting better, IMHO.) So I am curious if you tried simply writing your own code to handle all of the specific logic you wanted to accomplish? To me, the beauty of the platform is that it allows developers to write their own code without having to get into the weeds of setting up and maintaining a custom Linux based solution. At the same time, I completely respect that some folks want that level of control over their platform. My son runs Arch Linux as his main desktop operating system and writes embedded system code for a living. If he were ever to get into home automation in a serious manner, I would suspect he'd look long and hard at Home Assistant as well...and then eventually he'd probably decide to write his own home automation platform from scratch instead! :stuck_out_tongue_winking_eye:

Again, thank you for sharing your thoughts. You raise a lot of good points that may be useful to others considering what platform they'd like. Good luck and have fun!


#6

Just look at the AmpliFi forums. A friend of mine got the AmpliFi-HD and super basic things that users on even the cheapest of routers have, Ubiquiti has never bothered to put into the firmware for the Amplifi-HD. Things like QoS for ethernet connected devices, nope. IPv6 DNS settings? Nope.

As for HASS itself, I've always had an instance running for things that I could never get setup (or setup nicely) in SmartThings. I still use HASS for configuring my outdoor Pixel LED lighting (along with NodeRed and MQTT) and some one off Sonoffs that use MQTT. However, what has ALWAYS turned me off from HASS was the Zigbee instability. I still follow the Zigpy/Bellows Github (Issues · zigpy/bellows · GitHub) and they are STILL wrangling a bug where Zigbee devices become unresponsive after a few hours. It's been going on for almost a year now. That's what really brought the WAF to zero; When she gets up in the morning and the lights don't come on automatically, she gets, um, cranky.

Regarding WiFi on the HE hub, I'm VERY glad they don't have this. WiFi and Zigbee operate in the same 2.4ghz spectrum. Most "typical" users don't know this and when their Zigbee devices fail, the first thing they blame is the hub (or the device) without understanding channel overlaps in the 2.4ghz spectrum and how to avoid them. Having a WiFi radio right next to a Zigbee radio is a recipe for disaster, imho and I've seen that on HASS running on a Pi using WiFi with a Nortec ZWave/Zigbee stick.

While the OP has valid points and I respect and completely understand the programmer view, I don't see HASS as a replacement for HE, but more as a compliment to it. Especially not with the ability to run custom apps and drivers on HE without having to restart the hub after nearly every change.

@soumya92 Good luck and thanks for the good points and excellent write-up.


#7

Interesting write up and comparison of HASS and HE. I would have to disagree with you on many points though. HASS is 100% community driven meaning that if there is an issue with any of the peripheral “apps” such as location services then you rely on the community to fix. Same model as HE. Locative is a great example where too many changes by both the HASS dev teams and also independent developers caused too many rewrites resulting in a failed startup (I am sure there is more than just those issues but I followed that thread for a while). Also HASS is significantly more vulnerable for security and stability due to YAML and the perceived notion that there is a firewall installed (which there is not on initial installation).

I also disagree that HASS is anything but very complicated and is not very convenient for tweaking and modifying without really screwing things up. YAML is like an overly sensitive toddler. Have an extra space you can’t see and the system cries fowl without any notification on where you screwed up.

When HE was having some growth pains between 1.4-1.7 I was completely fed up with my Kwikset locks not working and had it in my mind to move to HASS blaming HE. After failed attempts and even sending equipment off I had no luck resolving my lock status issue. I spent 2 weeks setting HASS up watching Bruh and Dr. Zzz videos. Had nothing but problems with Zwave and Zigbee devices joining and working correctly. Eventually I had it running half the house when the I decided to move the locks over. To my surprise they didn’t work reliably with HASS either. So I bought Yale locks and within 4 hours had my house back to HE and notified HE staff that the KwikSet 888 are compatible but have a weak antenna and firmware for status reporting.

During those two weeks I also fooled around with Presence options and just like HE nothing was reliable unless you were using multiple solutions (BT/WiFi/GPS) and some of those solutions required a client. The community even recognizes the same limitations of WiFi and Apple location integration issues due to iOS sleeping and not being a HASS platform issue. Same solutions as here and same issues as here on HE.

Over all though I agree with you that there are improvements that need to happen. I miss WebCoRE and the simplicity and complexity you can achieve in pistons over RM. BUT Bruce has been listening to the community and actively addressing the shortcomings everyone sees and implements new functionality almost on every update; global variables is a great example. Remember HE is only a year old. HASS is three years old and has gone through (3) core rewrites with another one for the UI just being released.

Wish you the best but I know we will see you back as I suspect you are like the rest of us and keep your eye watching the solutions that interest you the most and one day (probably sooner than later) the issues you addressed above will be resolved either by HE staff or the community.

Thank you for your write up!


#8

Their (Ubiquiti) forums are depressing, especially their feature request section.

I would hope that HASS wouldn’t run into the same issues as other Ubiquiti products. It is open source still after all. But I think their goal is to roll a HASS hardware platform to tie into their EoT light and camera systems.

My Unifi setup (USG Pro4, USW-16XG USW-16 150POE, AC HD AP) while good, is over priced crap compared to another setup I run (10GB pfSense box, Mikrotik CRS 317, TP-Link SG3216, ASUS RT-AC68 in AP mode). Price to price and pound for pound capabilities the cheaper ham sandwich setup trumps the Unifi one.

Ubiquiti brought Chris Buechler on to help with the USG, which helped, but it’s still crap and meaningful updates just aren’t happening anymore. I can’t recommend Ubiquiti stuff outside of their WISP offerings, Edge series, and Unifi APs.

The latest Docker revision/update for HASS was scheduled before Ubiquiti purchased it. What happens next will be interesting.


#9

Good luck, a small reading here, HA in second place


#10

Excellent post.
Captures most of the reasons I want a HA setup.
Every time I dive into it I really don't make as much progress as I feel I should.
It's definitely more frustrating to set up but maybe more robust in the end.

I think I need to reset my setup to a fresh install and start again.
It wouldn't even automatically discover Tasmota devices as I hear it is supposed to.


#11

Very competent and apt review.

It's clear you are a bit more than the average consumer, and for that I'm sad to see you go, rather than seeing you committing to the excellent community by putting together some of the missing functionality and sharing it. That said, I get that many of your points cannot be overcome with ingenuity alone.


#12

A very excellent summation of the state of things by the OP... I certainly agree wholeheartedly on several points, like restricting verbose logs to support staff, limitations on backup and restore, and the sluggish UI..

That said, Hubitat and Home Assistant are two completely different beasts. Hubitat is a (relatively) mass-market product, while Home Assistant is targeted towards IT geek/tinkerer market. Hass.io attempts to at least simplify the installation, but the actual configuration is still cumbersome and complex.

Back when I was having issues with Hubitat Zigbee I spun up Home Assistant, first on a Mac, then a PI, then back to the Mac as I couldn't get Zigbee to work on the Pi. It literally took me 5 minutes to pair a SmartPlug and create the entities, and I still didn't have power reporting working. Yaml is the biggest strength and weakness due to its complexity and sensitivity to syntax. At the end of my experiment, I concluded it would take me months to rebuild my 400+ device system on Home Assistant.

That's not to say Home Assistant is bad or inferior... It depends on the use-case and what the intended goals of Home Automation.. For me, it's all about achieving the longest maintenance free operating period. I have 3 months out of a calendar year (Jan-March) in which I have enough downtime to work on pet projects. RV travel season starts in mid-April, and we open our cabin in May. From April-November we travel to one or the other, each and every weekend. During that time home automation becomes an even more critical system, but has to work as an appliance that should not require any care and feeding outside of changing batteries.

I do hope the Hubitat teams take note of your feedback as it contains some excellent suggestions to make Hubitat even better than it is.

Good luck with Home Assistant!


#13

I spent almost 4 years with HA. One of the reasons I left was the constant underlying changes. Originally there was an update every week. There were always breaking changes. Sometimes it was just a wording change and you could do a global search and replace. But other times it was a change in how something functioned or structural changes or how Zwave devices were handled and it could turn into hours and hours or even days of fixing.
In the past year, it has become a bi-weekly update schedule but there are ALWAYS breaking changes. Some updates don't have changes that apply to your setup but most do. And there is no benefit to skipping updates. You still need to go through all the updates you skipped to check and fix all the breaking changes within the missed updates.
The most recent change to HA has been most disturbing from a "stand back and observe" perspective. Support from the community has always been crucial to the success of HA. As in the forums here, it is where the collective knowledge and understanding lies. Over the last year and a half, looking at the senior forum members participating in the community forums, their numbers have dropped off significantly. Really significantly.
Today all the activity is in Discord chat. Sadly the collective knowledge and understanding scrolls off the screen and is lost.
Today I am pleased to have a supported platform. I may not have everything in the world combined into one hub what I do have is manageable.


#14

That is the one thing I always tell people when they ask me about home automation recommendations. There is no one hub solution. There might be somewhere down the road, but today, there just isn't.


#15

Interesting write-up. and I can definitely see why you might prefer Home Assistant! I actually came here from the other direction: switching from Home Assistant to Hubitat. Home Assistant was a lot of work to set up and maintain (I used Hassbian; Hass.io would at least save me from becoming a part-time Linux sysadmin but didn't support everything I needed at the time). As someone else mentioned, some updates also introduced breaking changes for me, so a careful read of the (long) release notes is necessary. You also say it works with more devices, but that really depends on what devices you're talking about. I'd say that Hubitat is mostly targeted towards Z-Wave and Zigbee users with the occasional LAN device (Lutron, Hue, etc.) or cloud integration. Home Assistant works well enough with most Z-Wave devices but has at least two ways to do Zigbee/ZHA, and the one I tried (ZHA component/bellows) did not support nearly as many devices as Hubitat, and unlike Hubitat, there was no way to import a custom "handler" for unknown devices. (This was also true for Z-Wave, but being more standardized, it would at least communicate properly with them and I could use templates to fix weird behavior.) Home Assistant certainly has an impressive list of other integrations/components, though I'm not sure Hubitat is targeting that market.

I was quite happy when I switched to Hubitat because it worked with all of my Zigbee and Z-Wave devices (except a couple where support was added; I got one of the first hubs very soon after launch). I actually kept a lot of automation on Home Assistant, and if I ever decided to go back to Home Assistant, I'd likely keep using Hubitat as a Z-Wave or (at least) Zigbee radio since it's so much better at that. All my logic has been migrated to Hubitat now, though. For many of my automations, I resorted to AppDaemon over YAML, but in either case, you might be interested in custom code for Hubitat over Rule Machine (I agree the UI can be a bit cumbersome)--not sure if you tried, but it's about equally powerful.

I actually still kept Home Assistant around, mostly for the UI and the cute, short-term device history graph it provides by default. Won't be sad if I lose it, and there are ways I could use it more to supplement Hubitat if I wanted, but I'm quite happy as-is. I can see where you are coming from, however, and wish you luck!

PS - If your hub is for sale, I think several people here may be interested in buying another. :slight_smile:


#16

Sorry to see you go. Have you taken a look at Homeseer. I use it in conjunction with HE for the things HE can't do right now. Some devices HE can't control and they use plugins, but they cost. More for the programmer like yourself. If you can write your own plugins that might be the way to go.


#17

My Arlo camera was "hacked" with Home Assistant. I followed old YouTube videos for an install. HA just updated so no new videos existed. I followed the video and improvising when things didn't match. The security steps were out of date but I thought I did a well enough job. Apparently I didn't and they gained control of my camera. I only added just the camera to HA and nothing else.

I powered down HA and had to change my Arlo password. They couldn't change the password cause I don't use the same password twice so the Arlo email password was different. But lesson learned, HA is not a safe platform to tinker. You have to know what you're doing or you run risks.


#18

Ah, I forgot to mention that the Hubitat community is awesome :slight_smile:

A few things to clarify:

  • All the SSL / authentication stuff is used when accessing my hub (HA or HE) from the LAN or through VPN. There is no remote access. SSL because otherwise Chrome and Firefox will annoy the crap out of me with their "insecure website! don't put in a password!" warnings. Auth because I don't want to manage subnets for guests (chromecasts get too complicated). Yes there are still some devices they could tinker with directly (Wemo, for example), but for the most part that limits damage from adding a compromised device to the network.

  • 5 GHz WiFi has existed for a while, and I try to use it exclusively, so Zigbee interference should not be an issue, while still giving me the freedom to move the hub around (for pairing or re-arranging furniture).

  • And finally, my hub is not for sale, because as pointed out, I'm still watching HE and might be back if HA turns out to be too much or if HE improves significantly. That, and I'm intrigued by many references to using multiple hubs (most with smart things, but HE should work too), using one as a proxy for zigbee/zwave from HA. It even has the nortek stick that I could use directly, if necessary.


#19

I think that’s more of an annoyance than an actual issue if it’s all local or through VPN.

I run so much stuff internal only, or only allow specific outbound traffic, that browser warnings are just something I have to deal with. You can always manually verify they are kosher every once and a while.

Besides having external verification via LetsEncrypt and feeling safe, not too much gain. I’d rather keep and know that my HE hub is talking only to the HE cloud via 8883 and 443 and that’s it. These days I like to just keep things dead simple rather than have complexity hidden behind simplicity. That’s probably why that other user wound up in the Arlo camera situation.


#20

I too have started to increasingly use HomeAssistant - and it's obviously a very different product to HE. They have made large changes to their UI options recently that make it look far sexier, and it has been mainly solid on an NUC. But Z-Wave and inparticular ZigBee are troublesome and many are using other hardware hubs for this functionality connected back to HA e.g. Zigbee2MQTT or DeConz.

But I am noticing that Hubitat is becoming a standalone peripheral to my system rather than the centre. In thinking over my journey I think now this is mainly because of the lack of decent MQTT integration provided by HE as that is central to my home now. But It really is a labour of love using MQTT within HE. We need a well integrated MQTT client.

I was thinking back over my HA frustrations during the last 20 years and just realised that now for the first time almost everything is do-able and does mostly work. I have more tasks now that just require time and attention to complete - and not tech advances or decent products. I feel HA 'arrived' during last year - but still only for the techies. A vehicle for mass adoption still eludes.