I want to turn off parallelism while sweeping, or find another way so I don’t get this error from deep mind control suite environments.
RuntimeError: Cannot make context <dm_control._render.glfw_renderer.GLFWContext object at 0x793b77fdd820> current on thread <Thread(Thread-5 (_run_job), started 133296402990784)>: this context is already current on another thread <Thread(Thread-2 (_run_job), stopped 133296824518336)>. X Error of failed request: BadAccess (attempt to access private resource denied) Major opcode of failed request: 152 (GLX) Minor opcode of failed request: 5 (X_GLXMakeCurrent) Serial number of failed request: 194 Current serial number in output stream: 194
Hello @b5wttv7orxtqz5c2-dlc , thank you for writing in and happy to help. Could you please share the following to better understand the inquiry:
-Code Snippet
-Sweep Config
-Environment Set Up – are you using Jupyter ntbk, etc?
Hi, so I fixed the error, though not fully understanding the exact cause, the agent script was a script which was submitted using subprocess.run from the previous script that submitted the sweep. My guess would be that some variables would have been persisted into the job submission (SLURM sbatch) and SLURM persists some variables into the script that causes wandb to not be able to connect.
Hello @b5wttv7orxtqz5c2-dlc , glad to hear that it was already fixed from your end. May we for the following:
-Code Snippet
-how you solved it
-what case/code works and not works
Hi @b5wttv7orxtqz5c2-dlc , since we have not heard back from you we are going to close this request. If you would like to re-open the conversation, please let us know!