With the new change to prevent test overruns, tests that are slow to complete result in an ensemble taking a long time to finish. There are potentially multiple causes of this, and it would be helpful to have some additional insight into what is taking a long time. For example:
Make it easy to list outstanding tests
Show whether the tests come from a single (failed) agent, etc.
Where feasible, we could also aim to eliminate the cause of slow tests.
With the new change to prevent test overruns, tests that are slow to complete result in an ensemble taking a long time to finish. There are potentially multiple causes of this, and it would be helpful to have some additional insight into what is taking a long time. For example:
Where feasible, we could also aim to eliminate the cause of slow tests.