I think I know what hapend (?). Basically if I use your tecquique AND keep zerotier client in a pi3 it does have problems with the pi3, its enough to uninstall zerotier one from the pi3 and UNCHECK the device in the zerotier network webui.
If , as I did, you DELETE the device from the zerotier network webui the pi3 disappears entirely from the network, do you know how to get it back?. Nothing I did works
I noticed that when you click DELETE you get this message
āThis marks a member as permanently hidden and banned. Deleted member records are purged from the database after a period of time. To āun-deleteā manually add the member.ā
For example when you start the Hassio Addon, it shows the address in the Log:
[12:22:04] INFO: ZeroTier node address: XXYY0011ZZ
You can probably see the same when you installed it manually on Linux. Just check the manual or logs when you start it.
Hi, never mind, the easiest and fastest solution is to delete the zerotier network and redo another one (I think zerotier has a small bug when you delete a member, because you simply canāt add it again).
From scratch to fully working it takes 5 minutes at the most.
Some new updatesā¦ since there is no Hassio Addon available so far for Face/Person Recognition via Camera i decided to first try to install Facebox via docker-compose and get it to run.
This means basically following this guide:
So i logged into my Ubuntu via SSH and created a new folder for facebox where i could add my docker-compose.yml for it:
mkdir facebox
cd facebox
nano docker-compose.yml
In the docker compose i added this:
Important to note:
In contrast to the official guide i do not have ${ YOURKEY } where the key goes.
This didnt work for me, only the plain key worked.
I added an additional environment attribute to load state (the faces i taught facebox) after every restart since facebox is stateless and does not save anything if your not a premium subscriber.
(This seems to only work from an web accessible address. I was not able to use filesystem for this.)
Since i already use port 8080 for the unifi addon, i had to switch the listening port to something different. i chose 8083 for it.
I had to use the _noavx version since my CPU seems to not be supported.
I chose the scan_interval of 10000 because of this post:
It makes a lot of sense to me, to only manually trigger the scanning. i.e.: When motion is detected.
I also used the python script robmarkcole posted here to automate the teaching:
For this i created a new facebox folder in my home assistant shared folder and for each person an extra folder with all the images. The script lies in the facebox folder and is simply called via:
python3 teach_facebox.py
After the teaching was done, i simply downloaded the state through the local website: localhost:8083 and added it as state.faceboy to my /config/www/facebox/state folder
Now, after every restart, it automatically load the state back into my facebox.
The automation for scanning and handling of the data i do via Node Red.
Forgive me mate, but iām really noob with home assistant. I successful installed facebox, and it works. I was looking around for re-teach faces on restart. How do you add the state downloaded from the facebox frontend as state for facebox?
Ok, i still have a problem. The state reloading only works if i manually restart the docker container of facebox, not if i restart the host machine. Why?
I finally got around to test this. And you are right.
I now implemented a new flow in my node-red, that upon restart just uploads the state file to my facebox.
Im using the simple URL method:
POST /facebox/state?url={url}
And then simply add the state file to my config/www folder to reference it in an absolute URL