latent bridge
Hey, so a couple of minutes ago, I bought a Zigbee gateway and I also found out that HA is the best Smart Home controller app. Therefore I was looking for a couple of sensors, lamps etc..
My question is, can I buy every smart product and implement it into Home Assistant, as long as it supports Zigbee? Or what exactly do I have to look after? Sorry for that question, but I really want to pick the right hardware for me.
wild sundial
The absolute best trick for zigbee is to check their ZHA/Z2M compatibility, depending on what you use!
Here is a link ot a compatibility repository
latent bridge
DAMN
tysm.. that helps me out
So as I have heard (I didnt install HA yet.. because I need to decide between Docker and VM) I can select between ZHA/Z2M right?
wild sundial
Correct, ZHA is integrated within HA, while Z2M is an addon that can be tacked onto HA.
Both work well, but Z2M allows for advance things such as custom 'quirks'
Quirsk being special zigbee configurations
latent bridge
mm i see, i hope i can get it somehow working (z2m in docker)
New To Home Assistant - Hardware Enough for HAOS or should I use Docker?
I have another question - I am using a DXP4800 with an Intel N100 and 16 GB DDR5 RAM, is this going to be enough for a HAOS VM? or should i use docker instead?
latent bridge
sorry for not clarifying, there is already an OS with Unraid on it
wild sundial
Some will tell you it's fine; I am in team : One host, one task
For most parts
Meaning that I have a dedicated NUC for HA, and a dedicated nuc for FRigate.nvr
latent bridge
hmm i see, yeah i was just thinking of using it with my nas because its already there
velvet mango
I started with a docker image, and quickly switched to a VM on my NAS. Updates are a lot simpler, installing plugins is simpler. It doesn't require a bare metal machine. I reserved 4gb ram for HAOS on my N100 NAS and it works great.
latent bridge
I started with a docker image, and quickly switched to a VM on my NAS. Updates a...
thats so cool, are you using anything else besides haos on that nas?
velvet mango
thats so cool, are you using anything else besides haos on that nas?
Adguard Home, Jellyfin, Transmission, Portainer on Docker and Tailscale directly on the OS.
latent bridge
ahh i see... i think i can run it as a vm too then
velvet mango
ahh i see... i think i can run it as a vm too then
I do have 32 gb RAM on that device though, although 16 would also be enough.
latent bridge
yeah unfortunately the ram prices just skyrocketed😭
velvet mango
This is my "idle" state with all the stuff I mentioned.
ruby ether
top -co%MEM and docker stats would probably be more useful.
velvet mango
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
912920 libvirt+ 20 0 6648836 4.1g 18440 S 13.0 13.1 60,42 /usr/bin/qemu-system-x86_64 -name guest=608e6ebb-b3e5-4776-98b2-2aeb1e4c015b,debug-threads=on -S -object {"qom-type":"secret","id"+
4788 root 20 0 262.5g 547356 157680 S 0.0 1.7 42:22.92 /jellyfin/jellyfin
7625 root 20 0 1471484 255332 10844 S 0.0 0.8 125:40.43 /opt/adguardhome/AdGuardHome --no-check-update -c /opt/adguardhome/conf/AdGuardHome.yaml -w /opt/adguardhome/work
693 root 20 0 1387516 140548 15524 S 0.3 0.4 27:34.01 /usr/sbin/tailscaled --state=/var/lib/tailscale/tailscaled.state --socket=/run/tailscale/tailscaled.sock --port=41641
4371 root 20 0 3613804 105300 42148 S 1.3 0.3 7,45 /usr/bin/dockerd -H fd:// --containerd=/run/containerd/containerd.sock
CONTAINER ID NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS
7486b2a61824 Nginx-Proxy-Manager 0.03% 124.5MiB / 31.1GiB 0.39% 4.73GB / 3.53GB 98.4MB / 2.41GB 27
a8e99b0b8d5e portainer 0.00% 40.96MiB / 31.1GiB 0.13% 393MB / 111MB 52.6MB / 4.33GB 8
e56287062fea transmission-openvpn-transmission-openvpn-1 0.07% 31.29MiB / 31.1GiB 0.10% 1.21MB / 506kB 20.1MB / 1.29MB 7
4d2ad1821878 AdGuard 0.02% 265.4MiB / 2GiB 12.96% 0B / 0B 170MB / 1.69GB 12
288ae61e3cf9 jellyfin-app-1 3.73% 2.149GiB / 31.1GiB 6.91% 187MB / 106GB 312GB / 1.24GB 2
The top process is the VM for HAOS.
ruby ether
This looks much more realistic. The 9G used made me suspicious. Im guessing glances (or top -co%MEM after entering login) inside HAOS shows it use <= 2G memory. 4G is a good ceiling though.
ruby ether
WHy a VM?
Because virtualization is awesome: https://gist.github.com/Impact123/c23c36eafe1672ec056233e450a86ae2
velvet mango
Yeah, the VM reserves that memory while docker just uses what it needs.
wild sundial
Because virtualization is awesome: <https://gist.github.com/Impact123/c23c36eafe...
I get all of this, but when you're chasing them femtoseconds for reaction-time, you can't add layers of complexity.
Plus the power of these nucs, mostly 4 cores with no HT, is nowhere near what I would recommend for a proper homelab with proxmox.
But again, I want my install to be as reactive as possible. I did see a difference when going from my Proxmox install to a dedicated nuc.
And the proxmox host is way above what the NUC I got for HA runs.
NUC:
Part Description Usage
CPU Ryzen 5560U
RAM 16GB
USB1 Sonoff Zigbee 3.0 Stick Zigbee devices
USB2 Zooz Zwave 800 Zwave devices
SSD1 500GB NVMe Operating Systems
UPS 950VA
Proxmox host:
Part Description Usage
CPU Ryzen 3900X
RAM 128Gb
GPU1 Nvidia P4000 Jellyfin Transcoding
GPU2 Nvidia P4000 Jellyfin Transcoding
GPU3 GTX 3090 Machine Learning
HDDs 8*8TB Plex Storage
SSD1 2TB PCIe3 M.2 Operating Systems
SSD2 500GB PCIe3 M.2 Incoming Download Cache
UPS 1500VA
gaunt remnant
if i may throw some salt on it, i (=me) would go for full haos vm running on ProxmoxVE