-
Notifications
You must be signed in to change notification settings - Fork 224
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
MultiprocessVectorEnv blocked forever #389
Comments
I think I've not experienced it. Can you provide more details? |
It happened three times so far. In all cases the system was under a heavy load and I was running more than one agent concurrently. The blocking agent was using at least 8 environments and all training was using CPU only. If you need more specific details don't hesitate. |
I have no idea how it can happen. Can you provide code that can reproduce this issue? |
Unfortunately, I can't. In addition it is not reproducible in a deterministic fashion. |
Update: it also happened when training a single agent with 16 envs on a 24-core CPU. |
Hi, I've noticed that sometimes, when running more than one multiprocessed agent the program is blocked forever because of the line
results = [remote.recv() for remote in self.remotes]
underMultiprocessVectorEnv
. Do you have any workaround for that?The text was updated successfully, but these errors were encountered: