I spotted that the device cannot work even by cutting off power.
It's probably that the device's firmware was corrupted by wrong flash writing. But I'm not sure, there is also small possibility that the device is not burned successfully.
Interoperation tests requires at two kinds of devices. Devices configurations should include device type. And when setting up test-bed, which device should be used as DUT should be configurable.
Currently GRL thread harness will not abort the testing even if exceptions are raised. We should reduce the trial times and shorten the delay between retrials.
In slow computer, harness service may not start in several seconds, a more accurate way is needed to verify that harness service is started before opening chrome to do test.