Is there a reason to run hassio over standard HA? I run standard HA on Ubuntu 18.04 on a Hyper-V server as well, but just regular HA. It works the exact same way as it would on proper ubuntu and no need for scripts or workarounds.
But I have never used Hassio so I can’t really tell the difference other than that it is (apparently less open and harder to install custom stuff), but has the ability of installing addons which standard HA doesn’t (though I know it is possible to get them running anyway)
That is why I switched to ESXi. If your hardware is supported (most NUCs are, although not officially) it installs in minutes. You can then manage your VM’s from a browser and they have access to hardware such as USB devices. I never used gpio on the RPI because it don’t feel it is very suitable for home automation use.
Actually I believe there are three flavours: Home Assistant, Hassio on Standard Linux and Hassio on HassOs.
For me Hassio on Linux is the best of both worlds, so that is what I use. It has the addons, but is just as open as “standard” Home Assistant. The script for the install on Ubuntu Server 18.04 LTM makes it quite simple to get running.
I’m curious as to why you feel the pi gpio is not very suitable?
My plan is to have it monitor 3 doors sensors and 3 motions from an unused alarm system, and make those statuses available to HA and push state variables to the ISY. I’ve never used gpio on a pi, so don’t really know what to expect.
I’m only using Hyper-V because I have an existing Hyper-V server with 2 Linux boxen, and a Windows server. I do wish Hyper-V would allow some hardware passthrough to the guests.
My system works with KNX i/o modules which makes it robust I also retain basic functionality should Home Assistant ever go down… If you are to use rpi gpio I would suggest to use at least some sort of rpi hat which gives electrical isolation. I used the Iono Pi for a while and was very pleased with it. Not using it anymore because it was underpowered to run my Hassio setup.
I was in the same situation but bit the bullet and migrated everything over to esxi. So far no regrets, VM’s are just as fast (maybe even faster) and very stable. The web-based management also is very convenient.
Yep, that’s what recently worked for me too on a 2012R2 Hyper-V Host.
Download the VHDX from the Hass.io link on @allan’s post above. Extract it to wherever you want to keep the VM’s disk. Rename as required. You might want to pre-create the folder tree for the VM.
Create a new virtual machine. Choose to put it in the path you’ve already set up.
Name it, select Generation 2, and give it some memory (I just went with 1024).
Choose to use an existing disk - the one you downloaded in step 1.
Before booting the VM, turn off Secure Boot in its settings.
Boot it up - the drive seems to unpack / set up to around 6GB but I’ve not really started giving the VM any jobs yet so it might grow further. I expanded it to 10GB for good measure.
Edit: I was having a peculiar problem when I tried to build two like this - and only the first one worked. Turned out to be a browser issue, I must have had some cached stuff from some earlier testing. Clearing my cache sorted access to the second one. I expect one could spawn as many as needed.
I’m thinking of moving to a Hyper-V VM (existing server) from an RPi3+, noting that I’ll lose USB passthrough for my UZB1 Z-Wave stick. So that’ll mean I have to buy a Vera for local z-wave control (AFAIK Wink hub requires the cloud?) or another way to get Z-Wave over TCP/IP?
The RPi3+ has started to give me complaints with power supply. I think I need to buy a higher amperage 5v supply (using a 5v 2.4A phone charger right now). Also a big headache trying to pair my Schlage locks to the UZB1 on the RPi, but my old Wink hub was a breeze… but that’s for another thread!
I’m thinking of moving to a Hyper-V VM (existing server) from an RPi3+, noting that I’ll lose USB passthrough for my UZB1 Z-Wave stick. So that’ll mean I have to buy a Vera for local z-wave control (AFAIK Wink hub requires the cloud?) or another way to get Z-Wave over TCP/IP?
I’ve done this before following these instructions as a guide (for OpenHAB). I was going to see if I can get it working with the VM image for Hyper-V, but I cannot figure out the login username and password in the shell to see what I can actually install.
It turns out that getting it into an addon isn’t hard at all. However, I can’t get the device to show up outside of the addon’s Docker container (it shows up just fine in /dev in the addon’s container, but not anywhere else).
Earlier this month, I hopped on discord to see if I could get some help there. @pvizeli suggested that shared memory support might be needed in order for this to work. I think that meant that I’d need to patch Hassio to allows a new privledged option to config.json. Given that socat ends up creating a pseudo-tty, I was also wondering if Docker’s SYS_TTY_CONFIGcapability might also help.
I’ve never used Docker and I’m very new to Home Assistant as well, so I’m in unfamiliar territory. At this point, I’m stumped and I’m not sure what the next step should be, but wanted to share in case someone who is more knowledgeable than I am wanted to chime in or pick up this work.
Have veru good expirience with HyperV with Hass.io, after pretty bad with Virtual box.
HyperV just works, starts with Windows, doesn’t need any maintenance.
Easy to setup… Real treat, not something you expect from MS :).
Really excited to try Home Assistant but equally frustrated trying to simply run it using the Hyper-V disk
Issue Summary
Receiving the following starting my VM using the hassos_ova-3.7.vhdx:
udevd[7]: starting version 3.2.8
udevd[8]: starting eudev-3.2.8
hv_baloon: Max. dynamic memory size: 1048576 MB
Uhhuh. NMI received for unknown reason 00 on CPU 0.
Do you have a strange power saving mode enabled?
Dazed and confused, but trying to continue
Uhhuh. NMI received for unknown reason 00 on CPU 1.
Do you have a strange power saving mode enabled?
Dazed and confused, but trying to continue
Details
I’m new to Hyper-V (but I am familiar with VirtualBox)
Server: Dell 9020 (just updated to most recent BIOS)
New Win10 install - Windows Update says I’m up to date with updates
Network card: Intel® Ethernet Connection I217-LM - Under Power Management, I’ve unchecked: “Allow the computer to turn off this device to save power”
I ran into no issues following the described steps
When I started the VM, and all the commands started scrolling, no errors/failures displayed
My variations to Aves’ steps
Step #4 I store the VM at the default on my C: which is an SSD with about 200GB free space
Step #6 I specified 4096 MB of memory
Step #7 it says: “Connect it to the network you want to use it on.”, I chose the virtual switch I created with connection type external pointing to my Intel® Ethernet Connection I217-LM
I also added a 2nd processor (as indicated in my above Issue Summary)
I was able to expand the disk successfully (and use it, as well as the original disk in what I described in the above steps unsuccessfully)
At step #8 when I clicked Finish to create the new virtual hard disk I got: “The server encountered an error trying to create the virtual hard disk”, “Failed to convert the virtual disk. The system failed to convert … \scsi#disk&ven_msft&prod_virtual_disk#2&1f4adff…”
I also tried
Setting up NAT network for Hyper-V guests (I can’t add the link her due to this post being restricted to 2 links) and using it, it seemed to setup ok but when I used it in my virtual machine I got at least on failure when I started it (I don’t recall it) so I reverted what I created, I figured using my external connection with my I217-LM card is the most likely to work
I also tried
The older hassos_ova-2.4.vhdx disk, but not only got the same error but I also got a few FAILED messages when I started the VM
Honestly, I wish I had another machine, with another NIC to rule that out, this “seems” like it should be pretty straight forward as others have stated their relatively easy success
I’m very new to Home Assistant, and I had the same “issue” as you. Until I accidentally pressed enter with console in focus. Made me LOL @ myself.
Turns out that the boot process on console is not reaching login prompt, but when it’s sitting @ the “Uhhuh. NMI received for unknown reason 00 on CPU 0.” message, the server has started and the application is reachable on http://vm-ip:8123. Just click in the Hyper-V colnsole and press Enter, it’ll spawn login prompt.
Again, being a complete newbie here I’m interested in what you mean by accessing the vm via “vm-pi:8123”? I didn’t know my vm-ip and I tried to get it doing the following:
From PowerShell (as Admin): Get-VM | Select-Object -ExpandProperty NetworkAdapters | Select-Object VMName,IPAddresses - no IP address appeared
From the hassio console: hassio > dns info, I couldn’t find an IP that would resolve using port 8123
Knowing I can get to the HA login page, I pinged hassio but got what appears to be an IPv6 address as follows:
C:\Windows\system32>ping hassio
Pinging hassio.local [fe80::18b6:da41:8574:c7b0%3] with 32 bytes of data:
Reply from fe80::18b6:da41:8574:c7b0%3: time=1ms
Reply from fe80::18b6:da41:8574:c7b0%3: time<1ms
Reply from fe80::18b6:da41:8574:c7b0%3: time=1ms
Reply from fe80::18b6:da41:8574:c7b0%3: time<1ms
Ping statistics for fe80::18b6:da41:8574:c7b0%3:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 1ms, Average = 0ms
You’re welcome. 2 & 3 no idea without troubleshooting more, but I see you have been able to login. So this point is moot by now.
Just for your future reference, ping was one good way to go, just as good as the dhcp lease table. Next time just add the -4 switch to force icmp to use IPv4, “ping -4 hassio”. and -6 will force IPv6.
I’ve seen the “Dazed and Confused” message before too. I expect it’s to do with the processor architecture (or lack thereof); I haven’t delved too much because as noted the server has actually started. Maybe it’s a transient error that resolves itself.
I have an AMD and an Intel host and the same VMs run on either, though you can’t live migrate them between architectures. Maybe whatever Hyper-V is doing to abstract the CPU operations between the VM and the tin doesn’t behave quite like a “real” CPU and confuses it?
I am no Linux expert, hence my favoring the HassIO image, but the two processors might be confusing it too. Also I don’t know if it’s 64bit off hand. If it’s only 32 bit you’re only going to get a max of 4GB ram out of it. Mine’s running fine for now on 1GB but it’s one of those pet projects that grows so I’ll see what happens in the future…
