What is the best device to run Hassio longterm with out reboot?

Hi, I have been playing with the raspberry pi as a Hassio server for about 3 weeks. Last week I started having to reboot the pi every 30 minutes because I could not access it from the web interface. I was able to use putty to login, but no web interface.
I read an articles that there are issues with the SD card. So I decided to create a virtual box and install hassio. I followed the out line bellow to do this.

http://www.thesmarthomehookup.com/migrate-your-hass-io-from-a-raspberry-pi-to-a-windows-virtual-machine/

So far , using the same configuration it has been working for over 24hrs.

I would like to know what is the longest time Hassio has been running with out a reboot?
If want to go over a year without reboot do I need to use something other than a Pi?
I would like to start a side business doing home automation and love the Hasso interface and how I am able to integrate so many sensors and switches and other things, but I am concerned about the stability. Any suggestions?

IMHO linux is for 24/7 operation. But to have a safest way I’d rather use 2 installation. So just keep one running an be prepared to switch to the other in case of failure. So even updates will work in tandem.
Definitely a difficult situation if your (like mine) will be boxed somewhere and the access might be troublesome.
Consider to use the be SD card which has the least wear factor or a system with internal eMMC, which seems to last triple than SD card,

1 Like

I rarely reboot the server my HASS install runs on other than for updates (non-hass.io, running in Docker). But the point is you shouldn’t have to update outside of HomeAssistant updates, which i’d recommend doing somewhat regularly to 1) keep up to date with the latest and 2) to avoid having to sift through tons of breaking changes when you do inevitably need to update.

What should I use instead of a PI

My home assistant machine has been up 24 days 22 hours 55 minutes.

1 Like

The release cycle is 3 weeks. Admittedly you don’t have to update but the longer you put if off the more difficult it becomes keeping track of all the breaking changes.

The release cycle has nothing to do with rebooting.

It does if things go tits-up.

I think I had to move it, which involved unplugging. Media server up 117 days.

I run hassio on Proxmox in its own Debian VM. Never hangs, runs for months, I never had to reboot it because of a hang, usually its whole server reboot for something like hardware upgrade or dust cleaning. I did not keep track of the record uptime but it was well over 3 months for sure.

In general I do recommend this kind of setup on some home server, x86 cpu with virtualization. I have an old gaming rig turned into NAS/HomeServer but if you need to buy nee hardware it all depends on how DIY you want it. You can build something or buy some nettop, minipc. The one you probably heard of is NUC but its just a brand name for a nettop pc from intel. Gigabyte Brix is the same thing for cheaper. AtomicPi is also an option. Zotac has a lot of miniPCs. Etc.

Why Proxmox? Because its setup and forget. Only remember how you setup the host, all VMs and CTs can be backed up daily to a remote location (i use cloud synced folder). Burn the server, dump it in the river - buy a new one, install proxmox, restore from backups - 10-20 minutes of your time then a few more to decompress backups.

If you run more than just Hass then it also helps to intelligently allocate resources to services. I dont need all of my 8 cores and 32gihs for hass, so I only let it use 2 cores and 8 gigs. Memory leak? Messed it up? Other services keep working just fine, only this VM is affected.

I had other posts discussing this, you can search my profile for more discussions like this, including a long thread about how to make Pi not wear out its SD card. Basically you just need to manage how and where your logs are written to.

You can also boot Pi from an HDD.

If I had to set HA server up for someone else I’d also make a separate Watchdog CT, it would try to open Hass UI page now and then, and if it timesout - reboot Hass VM. Crude but may help. Worth making it send some warning if it happens too often though.

sorry cant link, from a phone

1 Like

So is the issue with the pi is the sd card. Do you think if I use an external HDD will solve the problem? Or a memory stick?

I run Hass.io on a old zotac mag HD-ND01-U w/Ubuntu 18.04 and docker. It’s almost a 10yr old nettop. Works a treat. A bit slow starting up from a fresh boot, take about 15 min. to settle down. After that thing are snappy enough for me.

The longest continuous uptime is about 2 months, As I update the os regularly. Definitely less troublesome running on a x86 system, at least for me.

My 2 cents FWIW

I’m approaching 120 days uptime on my NUC (Ubuntu) … and that wasn’t a reboot due to a crash. I do apply HASS updates regularly though so I guess HA is rebooting every couple of weeks.

I have hass.io running on an old Windows PC (under VirtualBox/Ubuntu) and on a NUC (under Ubuntu) and on a Pi 3B+. Microsoft forces some updates which include reboots and under some circumstances VirtualBox does not restart (I have not jumped through the hoops to run it as a service). The Pi sometimes has problems, but improved substantially when I bought the official power supply. The NUC hasn’t rebooted since I got it.

Why why why?

Because that’s all I had at the time. Which is why I got the Pi. Which is why I got the NUC. It worked fine on Windows, as it runs under a VM that is not running Windows.

Ive been having lots of problems trying to run hass.io with virtualbox+ubuntu server on a windows PC.
Im not very experienced with this at all.
If i restart the vm, i cant get hassio to work again. Do you have any suggestions on how to get it working after reboot?

Michael,
Your post is very informative. One question I have is what OS is the native server running. I have an Intel NUC with Windows 10 but not sure I want to run Proxmox on Windows10. Would I be better installing Ubuntu on the NUC and then Proxmox on Ubuntu or just go with Proxmox on the NUC with windows OS?

Proxmox IS the host OS. It’s level 1 supervisor, meaning it’s meant to be installed on bare metal. You install Proxmox OS, and then you use it’s Web UI to manage it, run VMs, containers, etc.

You should not run Proxmox on something else. In fact I’m pretty sure a lot of the things wont work if it’s virtualized. However it is possible to install Debian and turn it into Proxmox, because Proxmox is based on Debian. But there’s little reason to do so, it’s cleaner to just install it initially.

I would also advise against using Windows in any form for anything server related, unless you absolutely need Windows. Otherwise it’s a waste of resources. Running headless Debian takes like 200-300Mb of Ram and 0-1% CPU in idle. Windows 10 will use around 2 Gb of Ram and will constantly use CPU with it’s numerous background processes, try to look for updates, restart itself, unless you tweak and cut it all out. It also uses like 30Gb of disk space just to install Windows 10. Debian is what, 1 gig? A couple?

And if you were asking what OS Hass uses for it’s own installs, like the image they distribute for Raspberry Pis - it’s HassOS. It’s buildroot, basically they built their own very minimal linux distro for Hass as I understand it.

EDIT: That said, what CPU is your NUC? And what do you plan to run on it other than Hass? If it’s just hass and your NUC is one of the weaker ones, like dual core CPU or something, then Proxmox might be overkill. You might want to just install HassOS on it and be done with it :slight_smile:

Michael, thanks again for the help.

The NUC is i5-8259U so I’ll have 4 cores and 8Gb of RAM. I don’t have any immediate plans to run anything else, but eventually may run a media server off of it.

So just to clarify, proxmox is an OS that I also use to create the VM (Debian) where i install and run HAAS. Is that accurate?

Thanks,
Mark