Can you elaborate? How have you generated the test alerts in the first
place?
If they come from Prometheus, you would most likely want to have
Prometheus stop sending them (by removing the relevant rule if it was
just for testing or by reverting the monitored metric to normal values).
If you created alerts using some other way (API requests?) you should be
able to remove them using this method as well -- either just wait for
them to timeout (and stop sending them) or actively update it with an
updated endsAt attribute.
If this is a complete throw-away test setup, you may also choose to stop
all alertmanagers and empty your alertmanager data directories (which
drops all stateful data such as alerts and silences).
Just speculating though; to be more specific, I think we would need some
more details on your setup.
Kind regards,
Christian