How severe does this issue affect your experience of using Ray?
- Medium: It contributes to significant difficulty to complete my task, but I can work around it.
Hi.
So, I have a class that I’ve decorated with @ray.remote and through these classes, I have them do a bunch of work which I then collect once enough of them have returned.
working_refs = []
for k in range(500):
for j in range(config.num_actors):
working_refs.append(
workers[j].play_game.remote(
k, config.temperature, config.temperature_drop
)
)
num_trajectories = 0
num_states = 0
done = False
while len(working_refs):
# ray.wait will return the trajectories one at a time
done_refs, working_refs = ray.wait(working_refs)
trajectory = ray.get(done_refs[0])
if not done:
do a bunch of work on that trajectory
# if we have enough data now, set done = True
else:
# we have enough data,
_ = ray.get(working_refs) # to ensure that the remaining sent off tasks don't just sit there
# how do I instead do: ray.cancel(working_refs)
working_refs = []
How do I instead do: ray.cancel(working_refs) at the end rather than just having ray.get because I don’t actually need those results and it’s slowing down the code quite a bit to wait.