Instead of describing the pipeline in terms of sequential ops, this switches to assets - at least for the directly ping-related datasets (i.e. excluding the network). This allows using partitioning throughout the system.
After a lot of tinkering, the live tests seem to run through - although the most_likely_path step is only seen starting, not completing. Not sure what is going on...
Instead of describing the pipeline in terms of sequential ops, this switches to assets - at least for the directly ping-related datasets (i.e. excluding the network). This allows using partitioning throughout the system.