Unable to recover remote devices

:fire: Firewall mode not applied due to error

I changed my start.sh it 2 ways,

  1. I run wifi-connect in a loop, so that it keeps checking if there is a network connection
  2. I lowered the CPU frequency to 300MHz max, just to lower heat and power consumption

Now, I simply cannot get updated application images to load.

I get “Heartbeat not received” and is sitting at 84% download, but I’m still able to log into the Host OS.

Any ideas how I can get these back into good condition?

getting the CPU frequency back to full speed helped a lot, but I’m still struggling to make sure devices update their image before they start. :frowning:

Hi Jason,

So that I can understand what it is you’re trying to do, and what the issue is, I’ve a few questions:

  • what device type is this? (guessing a beaglebone, but which one?)
  • you have wifi-connect running - but why did you need to need to change it to run in a loop? What’s the connectivity use case?
  • when you say I'm still struggling to make sure devices update their image before they start do you mean the device is not downloading the latest release before the services start?

Have you tried using a different update strategy?

Phil

Hi @jkridner, were you able to get everything running? Could you help us understand what the issue was?

phil-d-wilson balena team
February 9

Hi Jason,

So that I can understand what it is you’re trying to do, and what the issue is, I’ve a few questions:

  • what device type is this? (guessing a beaglebone, but which one?)

BeagleBone Green Gateway

  • you have wifi-connect running - but why did you need to need to change it to run in a loop? What’s the connectivity use case?

Sometimes the devices are relocated. Sometimes Internet goes out and a temporary phone hotspot might be used.

  • when you say I'm still struggling to make sure devices update their image before they start do you mean the device is not downloading the latest release before the services start?

Yes. I had an issue where some devices didn’t get the update that added timeouts to wifi-connect and they would start wifi-connect before an upstream connection was made and therefore they wouldn’t never connect. It was a race condition resolved with multiple reboots.

Have you tried using a different update strategy?

Yes.

Hi again.

If you use the kill-then-download update strategy, I would expect that this would not happen. Can you let me know if you have tried that?

Would it be possible to give me a dockerfile/compose to try and recreate this? I have a BBGW that I could try and use. :slight_smile:

Thanks for the patience.
Phil

yes, switching to kill-then-download helped, but the issue i mentioned is that it sometimes required multiple reboots to recognize the change in the field.

Hi Jason,

wifi-connect is actually not intended to allow changes in the network. If you are needing this solution, I would recommend running nmcli directly and creating your own solution.

I would also invite you to create an issue, because you are not the first person who has asked for this functionality, and I do not see an open issue for it.