I’m curious how many others are restarting HA semi-regularly for non-config related events. I’m finding I can’t go more than 2 days or so before realizing some component isn’t working or logging. The website is always available, just the data isn’t fresh.
Since .33 I’ve been doing it more often and lately because of some error messages I’ve been working on with a few other users, it’s been at least twice a day.
Prior to that I could go four or five days as long as I wasn’t working on an automation/script/config. I’m sure someone will respond to this by saying you don’t HAVE to reboot to reload scripts and automations or UI/Group updates, but frankly, reloading never worked correctly since it was released and you always end up with weird groups or extra tab views so I just restart.
1 Like
sometimes it runs for days, before i have changed something in the yaml.
and then i restart.
i have never had trouble that i needed to restart without having edited anything (except the 2 times my sdcard was corrupted)
How do you know when your SD card is failing? What are the symptoms that you encountered? I suspect mine is failing too because my HA cannot last more than 1 day without rebooting.
The last time I restarted HA was 27th December. The last time I rebooted my PI was 9th December.
I use appDaemon for all my automations, so I don’t have to restart HA if I want to make a change.
1 Like
Personally I never had to restart HA for something not responding. Only when I changed some config.
Maybe it also depend which module/object you have in use. If the library have some bug it might have to be restarted to work again.
Which component stop being updated/responding?
It’s my very personal opinion, but starting from version 0.33 I have to restart my HASS more often (almost every day) if I want to see all my components working (btw, not so many components).
1 Like
both times when my sdcard was corrupted, the rpi didnt restart (or actually kept on restarting)
I restart HA daily or else I lose the web-interface, seemingly at random. Longest mine has run without restart is 3 days. I restart the Pi weekly usually after updating.
I think your problem might have to do with the Raspberry, instability problem are not rare on the Raspberry, often the blame is put on the power supply.
I run HASS in a docker container on a HP Gen8 with limitlessled, some ssh controlled devices, and several mqtt enable devices, no problem so far.
Did you already check in log if you can identify something?
Glad to see it’s not just me.
Once I get the server fully situated I will be transferring HASS over to it. That will free up the Pi for other projects 
I also run on an rPi, but I don’t see any indications of hardware failure.
For me components will stop updating, sometimes the API won’t respond etc… Things in the logs are mostly network timeouts here and there. No show stopping errors.
What are you going with for a server? I’d love to have @touliloup’s set up but not all of us can afford a Proliant server!
Running on a RPi2, use AppDaemon for all of my automations. The only time I reboot is if I need to change something on-boot for the Pi or if I am changing my configuration.yaml or group-view.yaml files.
I came across an Proliant DL380 G7 and have no idea what I am really doing with it. I currently just have Zoneminder and Glances running on it. It has plenty of power to spare so I figured I would migrate HASS over to it at least. I know nothing about full servers or their capabilities. I know it has 2 6-core processors, 32GB ram, (I added some) about 3 TB of disk space, and 4 ethernet ports. I have Ubuntu Server 16.04 on it right now.
I am open to suggestions.
1 Like
Last reboot of my PI was Jan 2nd, and I think that was when I installed .35. Now I do have to restart HA quite regularly, but that’s because I add a new group, or a new device. Other than that, it’s solid as a rock. I have about 37 actual devices in the form of wall switches, outlets, and lights. I have another probably 40 input_booleans and sliders used for wake up alarms for the kids and myself.
You can install hass on it (at best in a docker container), plex, duckdns, a git repo, gitlab, …
@rpitera I’m no Croesus either ^^ I got my Proliant Gen8 for 190€ and needed it for other purpose, it’s surely too much for just hass 
1 Like
I a not familiar with the docker version, I have only successfully run the virtualenv. What is the advantage?
The main advantage is that you can create a new container to try some new configuration/development.
If you’re not satisfied with the change or if there is a problem you can simply restart the previous container and go on with a fully functional hass.
And continue your dev/change later on the new container.
Also in case you reinstall the system, as long as you stored the configuration and Dockerfile (in a git report at best), you don’t need to reinstall python/pi/etc… docker will do everything automatically.