-
Notifications
You must be signed in to change notification settings - Fork 5.4k
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
[Bug] Can't pickle function objects error on ray.init #19938
Comments
Is it reproducible from conda env? (doesn't mean you should use the conda env, but I'd like to understand if it is the pyenv issue) |
I have no idea, I don't use conda |
I encountered the same problem, how did you solve it? |
I did not! Still looking for a solution and hoping Ray team will help |
This issue has been solved for Python 3.8.2rc1, Python 3.9.0 alpha 4 or late versions. |
@MXD6 Do you mind trying download |
My issue is that I need it to work on python 3.7 :( |
I will try it as soon as I can |
have you solved this problem? |
It did not resolve the issue:
|
I use conda env and encounter the same problem. |
The same problem occurs with conda env and Python 3.7.1 |
This seems to be an known issue with Tensorflow2. |
It looks like there are many people going through the issue. Can you try setting up the priority cc @ericl ? |
@Ramtin-Nouri are you seeing this error even when you didn't import tensorflow at all from the python script? |
@rkooo567 Yes. Even this would show the crash:
It seems as if only tensorflow 2 is installed its enough for the error to occur !? |
I ran into this problem after I decided to use pickle5 to check that my users aren't calling ray with too large of an argument list. My environment is ray 1.2.0, pickle5 0.0.12 Whether my code includes pickle5 or not, ray.init() fails. When I "pip uninstall pickle5", all is well. I don't have any tensorflow installed in this venv, so this pickle5 problem appears to be the fundamental problem. Presumably ray has a conditional import for it. |
Same problem with 3.7.0. I have no It might be a issue related to python. |
You might want to examine your stack trace, it would be surprising if pickle5 is mentioned there if it's not installed. The python version dependency might be that 3.7 backported some stuff in a middle version. The entire pickle5 module is unusual because it's a backport of the 3.8 pickle for earlier Python versions. |
Upgrading from 3.7.0 to 3.7.10 resolved this for me. Thanks! |
For me, I had this issue when I built Ray from sources. It was resolved when I re-create the virtual env |
I found this too. Ubuntu 20.04, python3.6.3, ray1.12.1. |
After reading this thread, it seems that it's not a problem with Ray and this issue could be closed. |
this method can solve my problem |
Search before asking
Ray Component
Ray Core
What happened + What you expected to happen
Running
ray.init()
throws this error:Versions / Dependencies
Ubuntu 20.04, Python 3.7.1 via pyenv, ray 1.7.1
Reproduction script
Anything else
Original discussion in Modin project: modin-project/modin#3601
Are you willing to submit a PR?
The text was updated successfully, but these errors were encountered: