Error in parsing supportedDeviceTypes (balena.yml) in balenaHub

balenaHub
1

Dear all,

just found an potential error within balenaHubs handling of supportedDeviceTypes:
Within my gpsTime fleet ( gpsTime/balena.yml at main · nmaas87/gpsTime · GitHub / balenaHub: an easier way to find and publish fleets and projects for edge devices ) I added following list:

  supportedDeviceTypes:
    - fincm3
    - raspberrypi4-64
    - raspberry-pi2
    - raspberrypi3
    - raspberrypi3-64
    - raspberrypi400-64
    - raspberrypicm4-ioboard

however, on the balenaHub page it only shows works with RPi 3 64 bit, RPi 4 64 bit, RPi 400 and RPi CM4 IO Board. If I am pressing the “Get started button”, I can only add a new device for these 4 device types. The Fin, RPi 2, RPi 3 32 bit are ignored. I have a guts feeling that it might be because these run 32 bit code? (however, the Fin should be able to use 64 bit with CM3 …)

I found a similar problem on balenaSound ( balena-sound/balena.yml at master · balenalabs/balena-sound · GitHub / balenaHub: an easier way to find and publish fleets and projects for edge devices )

  supportedDeviceTypes:
    - raspberry-pi
    - raspberry-pi2
    - raspberrypi3
    - raspberrypi4-64
    - fincm3
    - intel-nuc

however, all RPi Versions show up, but the intel-nuc is missing.

If you use the Get started, add a new device button you can search for other device types, but only the ones also shown in the “Works with” area will show up. Meaning, I am not able to push my gpsTime fleet to a Fin, RPi 2 or RPi 3 (32 bit) device.

Ant thoughts?

Cheers,

Nico

2

Sorry for pinging you @chrisys - but maybe you have an idea what I did wrong or if I actually broke something. Would not care so much about the issue, but got a presentation on my gpsTime fleet next saturday (not this one :wink: ) - and will also talk about balenaHub shortly as the means to install it - and it would be nice if I at least had an idea if I can get it working by then or just tell them RPi 2 support is coming … somewhen :slight_smile:

Cheers
Nico

3

Ok, I got an answer:
Fleets seem not to support yet multiple architectures.
I deleted my fleet (which was built as RPi 4) and readded it as a project (built for RPi 2) - now all options are available.

I guess its because of this: A project always needs to be created as its own fleet within the users own balenaCloud account. So you can basically choose which architecture you want, its independent.

Fleets however depend on the “mother fleet” and only support the machine type build with it. This does explain why balenasound works with all RPi but not with the Intel-NUC.

Handling a fleet as a project is a bit more of a hassle, as the user needs to create their own account, but its also more flexible if the user might want to set variables and you don’t want to include a webserver-gui to make it easily accessible.

4

@nmaas87 you got it, fleets are not yet multi-arch, although that is definitely something we’re working towards, and you got it exactly right with regard to projects too.

If there’s anything you think we can do to make this clearer in the meantime until we are able to roll out multi-arch fleets, let me know!

1 Like
5

Thanks @chrisys - works for me, know that I got it figured out. I guess its a little hard to describe and normally people have a tighter usecase / platforms arrangement. So most should not stumble upon this for the moment. But I see the complexity in getting this right, provided the nature of fleets and now the added usecases with balenaHub. Best of luck :slight_smile: