search

srcfl / srcful-gateway

Sourceful Energy Gateway: Connect your solar inverter or Utility meters to earn tokens.
https://sourceful.energy/
MIT License
5 stars 2 forks source link

RAK gw in Beta Fleet produces strange signature #151

Closed h0bb3 closed 7 months ago

h0bb3 commented 8 months ago

This looks odd - a plain RAK v2 should work out of the box or (i.e. no dual mining)?

image

h0bb3 commented 7 months ago

Seems to be working with the dual mining firmware and the rpi4 compose files.

DIY Dual mining FW 0.8.0 image

Rpi4 compose running on DIY Rak v2 image

h0bb3 commented 7 months ago

For that machine the crypto chip seems to be wonky:

image

I will try a reboot of it...

h0bb3 commented 7 months ago

remote reboot did not help - possibly a power cykle...

h0bb3 commented 7 months ago

Seems to be linked to #132

h0bb3 commented 7 months ago

After an unrelated docker rebuild on my Rak i suddenlty got this:

{"deviceName": "ATECC608", "serialNumber": "0123748154192e58ee", "publicKey": "00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000",

correct serial but wonky pubkey?

h0bb3 commented 7 months ago

Restarting containers did nothing but power cycling the device removed the problem....

h0bb3 commented 7 months ago

One idea could be to upgrade the cryptolib to some newer version. But it would be good with some repeatable scenario that triggers this. It seems to be connected to deploying new images, and/or restarting the containers?

h0bb3 commented 7 months ago

It would be interesting to know if the Nebra/Helium code can handle this situation. So it would be interesting to check this in the dual mining scenario.

h0bb3 commented 7 months ago

In the new threaded queue I tried reloading the first page that show the cryptochip info... got some really strange behavior in some cases it was ok, other we got the serial, other nothing, and also apparent crash and restart of the service:

image

will continue investigation....

h0bb3 commented 7 months ago

moha ha ha ha this was indeed a threading issue and the library needed to be protected by thread locks.

h0bb3 commented 7 months ago

This could potentially/probably be fixd by now. So I will close this as we will see in the next firmware release if the problem was removed.