Runs we could not finish

Recognise a run that a problem on our side stopped, know that it costs nothing and run it again.

Sometimes a problem on our side stops a run: a server of ours stops halfway, or a browser fails to start. That is ours to fix, so a run Rehearsal could not finish costs you nothing and says nothing about your app.

What Rehearsal does first Link to What Rehearsal does first

When a run stops moving for two minutes, or waits in the queue for two minutes without starting, Rehearsal starts it again by itself. The new attempt keeps the results the run already has and runs only the tests that are left, and it uses no extra test runs.

After three restarts, Rehearsal stops trying, and the run ends as Could not run.

What it costs Link to What it costs

Nothing. A run that ends as Could not run counts towards nothing on your plan, including the tests that passed before it stopped. It is never marked Failed, because the failure was ours.

Where you see it Link to Where you see it

WhereWhat it says
The run's page"Something on our side stopped this run. It counts towards nothing, so run it again when you are ready."
Runs"Rehearsal could not finish this run. It costs nothing."
A check on GitHub"Could not run these tests. You were not charged." The check asks for action and never passes, so a check you require keeps the merge blocked.
The CLIrehearsal run wait reports the verdict could_not_run and exits with 5.
Run emails"This run did not finish", with a line saying it was not counted.

Run it again Link to Run it again

When you are ready, press Run again on the run's page, then Run in the dialog, which needs the Manage projects permission. The new run checks the same test versions at the same addresses as the one that stopped. From a terminal, name the run by its number or its identifier:

Terminal
rehearsal run again 42

On GitHub, re-run the check instead. Either way the new run is an ordinary run, and its tests count like any other.

Next steps Link to Next steps