Need NUC advice (migrating from Pi3B

For my use case (Wired ethernet, Z-Wave using Aeotec USB, Mochad (X-10), Sonos, Envisalink Alarm integration, 7 day log purge, booted on SSD, off the top of my head at least 2-3X as fast as the SD card was (caveats–going off my memory here, HASS itself is faster than it was a year ago). I would say the performance increase was similar to going from hard-drive to SSD on a desktop computer.

Case in point, just checked and it takes <7 seconds to pull up full (7-day) activity Logbook (~3500 log entries) and < 2 seconds to pull up the History pane for same period.

Client computer is Macbook 2018 13" (4-core), Chrome browser.

My Z-wave is essentially, instant. I did use a 12 foot USB extender cable to mount the Aeotec USB on the ceiling of my basement–that definitely helped Z-wave range/speed.

1 Like

Well, logbook & history now even crash HASS completely (wanted to also time it knowing t would take more then few mins). DB is working fine as I get a history card on a device.

What do you see with “Free” command in terminal? I have almost 400mb of unused memory. Oh and I’m running pihole on this pi as well.

Size of .db is 2gig. memory & CPU below

Your mem and cpu utilization look fine. IMHO, you would almost certainly benefit heavily from an SSD vs SD card. FYI-My db is only about 1gb, about half of yours, with a 7-day retention period–even then, I am logging a bunch of activity that I don’t care about–I think I will exclude more activity than what I’m currently logging!

Note: I’m running regular 'ole HASS in a venv, not HASS.IO

I just performed a Vacuum on the the sqlite database and it decreased the database size by 90%! Viewing a full 7-day log now takes only 3 seconds! (Down from ~7 seconds). You should probably try optimizing your db with vacuum.

Was looking into this but does not work on hass.io.

I’ve just done something similar and offloaded most of my Home Assistant from an RPi 1 onto an old core 2 duo laptop that was gathering dust. I kept the RPi running with a stripped down HASSIO to run HomeKit and my GPIO setup. My setup is not huge (yet) but this laptop is idling at 2% utilisation. A modern NUC just for home Assistant would be overkill :joy:. Even the Pi is happy at 10% utilisation after I turned off virtually every component.

1 Like

I’m using a NUC5CPYH (Intel Celeron) with 4GB RAM and an SSD with hassio and a few other docker containers (Node-RED & LMS) under Ubuntu 18.04 server. Performance is excellent and I have had zero problems with it over the last 9 months, to the extent that I have wall mounted it above a kitchen cabinet and only reboot it after the occasional update. You should be able to pick one up for €150.

2 Likes

same here, except I have 8gb
I run my weather station from the same NUC, and these dockers
Home Assistant (not hasssio)
mysql
influxdb-grafana
MQTT
dockerman
portainer

I run on Linux Mint and performance is fantastic. I could want more, but I wasn’t willing to pay extra for a i3/i5 as I got this from ebay for about £70

Seems like a good idea, just ordered one (actually the NUC6CAYH which is slightly newer but approx. same price) Incl. 8gb mem.

Wanted to go for an I5 gen 8 so I could re-use for other stuff if it did not work out but getting close to 600eur and that is a bit steep for just an experiment and with just one sata port these type of NUC’s or also not ideal to host >5tb of data.

To follow up on this;

Got the stuff (NUC6CAYH & 8gb & 120gb ssd) in last Wednesday. Within 15 mins hardware was up and running. 30 mins later Ubuntu 18.04 + Hass.io was working. Just followed the steps on the hassio install page bottom part. Don’t forget to install docker ce (follow steps on link provided on above page).

Then restored my most recent snapshot, swapped my zwave usb stick and was fully up and running…

2 Likes

As I said piece of cake. How are you finding being to have the system back up from a reboot in less then 2 mins?

In all honesty I’ve never found booting a big issue. My Pi was generally up <1 min (or at least did not feel that long/too long). This boots significant faster and more important I can restart just hass without rebooting whole system.

What I still don’t get is how I can make this easier to ‘recover from disaster’ then the RPI. With the RPI I would of course have snapshots avail and now and then create a copy/image of the complete SD card so in case of disaster (either hardware or software) I could just use the img I had.

What is the best way to do something similar in the NUC setup assuming hardware is fine but I eg scew up my hassio. I mean next to the snapshot restore I would love to have something I can easily restore the complete thing.

This is an old thread but just in case you did not find a solution: Clone the drive with clonezilla.

1 Like

well if your hard drive fails you need to get a new hard drive: although in my 40 years as a user of IT I never had a hard drive fail on me (also because I changed them every 3/4 years or so).

If its not your hard drive, you may need to do a reinstall of your OS (it happened to me once with Ubuntu, and it took me 10 minutes to reinstall it)

1 Like

SSDs on the other hand… I’ve had one (of 5) fail in the first 12 months. Replaced by Samsung without any fuss.

I recently just built my own nuc. I used a j4105 asrock m-itx motherboard, 8 GB of crucial mem and a 120 GB crucial ssd. It has been running great.

i’m using a Intel Skull Canyon NUC with 16GB and very happy … in one year i went from RPI3/HASSIO ->RPI3/Debian->Intel NUC/Ubuntu16->Intel Skull Canyon NUC/Ubuntu18 … i’m now using 25 video cameras, tensorflow on 5, 5 HADashboards … don’t underestimate how much horsepower you will need … i did

1 Like

@netrunner did you install hassio or hassbian with the sd card + ssd setup?
Can you recommend a guide on how to boot raspi3b+ from sdcard but use the ssd for the rest of the os (to push the veavy writing to ssd)?

Interested in using such setup for mysql + influx, grafana, plus maybe motion eye for 2-3 cameras - raspberry pi 3b+ & SSD would be ok?