- One test that hangs the machine disrupts all following scheduled tests
- very high
Fuego needs to gracefully recover from this (which requires either manual intervention, or reboot of the hardware).
I have the capability for this, but I'm not sure where to put it. new: ov_hardware_reboot?
This is simple for a ttc-based client, but may be hard if automated power cycling is not possible.
Fuego should automatically do a hardware reboot, if the board health detection fails (in ov_transport_connect)
We need better error handling in ov_transport_connect - it ends up in signal_handler, when it should be issuing it's own error message and let the caller decide what to do.
pre_test should reboot the board if it can't connect. - No - that covers up lab problems.
NO, NO, NO - post_test should check for dead board, and reboot it then, with a massive failure message in the test that just ran!