Hello there,
is here someone who has been previously used OH instead of HA?
I am thinking about to move to the dark side… but I am afraid of stability of HA.
My current setup is RPi4 with 670 items (devices), 142 things on 27 addons. I am not a programmer, however, I am using great integration to VSC and writing code on RulesDSL, which I find OK.
Mostly my devices are Xiaomi, Tasmota, Shellys, SolarEdge inverter, IKEA, web services…
Too many devices from different suppliers… I found all the addons, but I am afraid of stability. Should I? I have heard, it is sometimes unstable.
HA is stable as hell. I’ve been running it for over 2 years now and never had any stability issues.
Though a ZigBee mesh can be unstable and mess up the mood. Also bad automations can take care of bad experience. But my experience is that it’s always human interference that causes trouble. So if you tend to keep stuff simple and clean it will work rock solid all the time.
Literally any junk machine is enabling you to test this out. If you’re able to schedule some downtime, you may only need a free SD card to test HA on the RPI. (the experience will be slow though, for better performance a hdd or an ssd is better.)
2 years ago I almost started in OH, there were some forum post stating that OH is more stable as HA is more aggressive in development.
Now I think HA has a good balance between progress and safety, plus the core staff is motivated financially as well to build a good product - which helps a lot.
Btw. 2 years ago I did choose HA because it seemed easier to integrate my alarm system - which was fairly easy and I’m a happy HA user since then.
HA can be a bit unstable in the beginning of the month if you are a first mover on updates.
There are alot of changes with each months update and of course there will be a few bugs too, but the team is incredible fast to correct them.
If you look at the updates for each months you will see that there will be an update around each second day the first third of the month, so wait with the update until the 10. day and then it will be stable.
I evaluated both HA and OH (and a couple others) when I first started. I don’t regret going with HA. The core itself has been rock-solid stable for me.
But… It’s not for the faint of heart. As mentioned, there are frequent core updates. These updates often require hands-on work to fix breaking changes, which are fairly well documented.
The biggest problem, in my experience, is that updates to integrations are rolled into the core updates. Integrations are notoriously hard for any one developer to test. They can only test in their own environment. Yours is probably different. Bugs are common, and not documented anywhere. It’s possible to update the core and then run into a bug in an integration, even if you diligently read the breaking changes list. The only non-invasive way to back out one of these integration updates is to back out the entire core update. In fairness, the developers usually do a great job of jumping on these bugs, but by the time one is fixed (in the next release) another could be introduced in a different integration you need. Once you get a couple of versions behind, catching up takes a lot of effort.
Right now the best a user can do is to minimize the number of integrations they depend on. I did put in a feature request to decouple the integration updates from the core updates, but it didn’t get much traction.
This shouldn’t be a show-stopper. Just go in with eyes open and understand that HA, and probably any such system you’re actively tinkering with, is going to require a commitment of your time to maintain. I just finished digging myself out of a three-releases-back hole and I’m still a fan.
I’m using HA for over 5 years now and can only confirm @frits1980’s statement. Rock solid, every issue/unstability I had was caused between the chair and the screen
I consider HA to be a particularly unstable platform, for this very reason. Frequent updates have breaking changes that, unless you have time to read all the documentation about the update, will break an integration, script or automation.
If you have the ability to run it as a container, then you can pin a particular version until you have time to read about the upgrade and sit down to iron out the breaking changes. This is what I usually do.
If you want an example of the havoc caused by breaking changes, read up on the z-wave and Tuya issues after the last major update (2021.10).
I would compare HA to Arch Linux more so than Debian.
HA have been great for me and so far no complain. Off course people have to remember regarding breaking changes and read the release document. Most people having problem due to just upgrade without know what they are getting into (its happen to me during my early adoptions)
Move into the light where developers actually care about user input. documentation is actually useful because it is written by the developers who create the code. openHAB expects users to write the documentation for the core. For addons OH requires the developers to provide documentation though.
The OH leader’s actions, many times, do not agree with his actions. I know for a fact the "stable OH 2.5.0 was released with a bug that was reported over 6 months previously. They refused to actually fix the problem in the core code but worked around the issue in 2.5.1 without changing the core. New users who installed OH & the addons package were unable to configure their system. I know this because I discussed it at length with the lead developer and pushed for a resolution.
A: Configure your recorder properly, and then you can leave HA alone, it will work alone.
B: IF you have new ideas (and you will have ideas), you have to choose:
=> B1: Realize those ideas in your home system => Your will cause unexpected downtimes, that affects other people in your house
=> B2: Realize those ideas on an sandbox system => Your will not cause unexpected downtimes, but you will have to spend more time to maintain your system
The problem with B2 is (in my experience): You have not two homes, one live-home and one sandbox-home .
I have done this on occasion to test breaking updates to integrations. If you have the capacity to run docker containers, it’s trivial to set up a new, test instance of HA.
My underlying storage is ZFS, so I usually clone my data and spin up a new updated container to upgrade. If all goes well, I get rid of the old container. If not, I get rid of the new one and wait for the problems to be ironed out.
In the last 13 months I have not had a single stability issue, though I am not running on an RPi… that seems to be the source of 90+% of stability issues judging by forum posts.
