Skip to content
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

zmq.error.ZMQError: Too many open files #69

Open
lbsswu opened this issue Jul 17, 2024 · 6 comments
Open

zmq.error.ZMQError: Too many open files #69

lbsswu opened this issue Jul 17, 2024 · 6 comments
Assignees

Comments

@lbsswu
Copy link

lbsswu commented Jul 17, 2024

I have tested serl in semulation for several times, the following errors are always met when executing "bash run_actor.sh". Could anyone help?

Pasted image 20240717152144

@youliangtan
Copy link
Member

Hmmm, does the actor process kill itself? and does this happen consistently? If you kill and restart the actor node (while keeping the learner running), will you still get this error?

@lbsswu
Copy link
Author

lbsswu commented Jul 18, 2024

thanks for your reply!

Hmmm, does the actor process kill itself? and does this happen consistently?

I think the actor process kill itself. This happen consistently.

If you kill and restart the actor node (while keeping the learner running), will you still get this error?

I tried several times in the same command window, and didn't notice this. I'll clear all the processes and have a try.

@lbsswu
Copy link
Author

lbsswu commented Jul 18, 2024

screenshot-20240718-192308

I killed all the processes and rerun the code, the actor still failed with the same error.

@youliangtan
Copy link
Member

Okay, I think this symptom seems to be due to the underlying replay buffer update from the actor node side to the learner node being slow. Thus it constantly make multiple requests yet stuck in that loop. The best way to resolve this is to try uncomment this line:

This experimental_pipeline_port=5547 uses a different communication protocol, which proves to be more robust to bad network.

Let me know again if this works. If so, I will try to migrate the code to use this by default

@lbsswu
Copy link
Author

lbsswu commented Jul 22, 2024

Okay, I think this symptom seems to be due to the underlying replay buffer update from the actor node side to the learner node being slow. Thus it constantly make multiple requests yet stuck in that loop. The best way to resolve this is to try uncomment this line:

This experimental_pipeline_port=5547 uses a different communication protocol, which proves to be more robust to bad network.

Let me know again if this works. If so, I will try to migrate the code to use this by default

Thanks! the actor is okay, but there is a new error on the learner side.
screenshot-20240722-093814

@WenzheHe
Copy link

WenzheHe commented Sep 6, 2024

I met the same error..

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants