How to use labgrid-client to flash image on remote pc?

[daocloud123]

When we use labgrid-client -p xxx fastboot or something similar, how could it works if we execute labgrid-client on a machine which not directly link the board?

If possible you could give me an example configure of exporter for such resource? And suggest me the full labgrid-client command example? Thanks!

Additional, does fastboot runs on the machine which execute labgrid-client or runs on the remote resource?

[Emantor]

When we use labgrid-client -p xxx fastboot or something similar, how could it works if we execute labgrid-client on a machine which not directly link the board?

labgrid copies the file to the machine where the board is connected to (the exporter) and than does an SSH onto the remote machine and executes fastboot with the copied file.

If possible you could give me an example configure of exporter for such resource? And suggest me the full labgrid-client command example? Thanks!

exporter-example-group:
  location: "In an example location"
  AndroidFastboot:
    match:
      '@ID_PATH': 'pci-000:07:00.0-usb-0:1.3:1.0'

The udev path above was taken from one of my exporters. If you have a fastboot device connected to the exporter, you can simply run labgrid-suggest to get the correct entry for the exporter YAML.

Additionally you'll want to look into the introductory videos from my colleague over on the pengutronix youtube channel.

Additional, does fastboot runs on the machine which execute labgrid-client or runs on the remote resource?

It runs on the remote machine.

[daocloud123]

This is really useful! A final question before I go to the youtube:

If possible I could configure the infrastructure as next?

1. machineA (directly link one device to this machine)
2. machineB (runs exporter on this machine)
3. machineC (runs labgrid-client)

Then on machineC, I run the labgrid-client, then labgrid somehow get the configure from machineB, and send images to machineA, finally runs fastboot on machineA?

In a word, the exporter not runs on the machine which directly link the board, is it possible?

More, if fastboot not support that, do we have other resource support such configure? Or this is not as design, and should be avoid for all kinds of resources?

[Bastian-Krause]

If possible I could configure the infrastructure as next?

1. machineA (directly link one device to this machine)
2. machineB (runs exporter on this machine)
3. machineC (runs labgrid-client)

Then on machineC, I run the labgrid-client, then labgrid somehow get the configure from machineB, and send images to machineA, finally runs fastboot on machineA?

In a word, the exporter not runs on the machine which directly link the board, is it possible?

No. The exporter exports (hence the name) its resources, so there is no point in running it on a different machine.

More, if fastboot not support that, do we have other resource support such configure? Or this is not as design, and should be avoid for all kinds of resources?

This applies to all resources, except for stand-alone network resources (i.e. resources with a host attribute). Have a look at labgrid's Resources Documentation for a list of all available resource types.
Also see the Architecture Overview to better understand what roles the different components play and how they interact.

[daocloud123]

Thank you, this answered my concern!