Segmentation fault (core dumped) about isaacgymenvs HOT 14 CLOSED

I met the same situations Segmentation fault, and all the words displayed in the window of Isaac Gym were too vague to distinguish. Have you solved it？

Sorry, @nicholas0717 . I haven't solved it. But I found the reason -- my graphic desktop was broken. Therefore, If you run issacgym on a headless ubuntu server, I recommand you to install ubuntu desktop by

sudo apt-get update
sudo apt-get install ubuntu-desktop
reboot

from isaacgymenvs.

I also found Segmentation fault (core dumped) happen if I launch isaacgym in tmux window (it works well without tmux). Just FYI.

from isaacgymenvs.

For anyone tries to install ubuntu desktop due to this issue, make sure there is no warning window.
Just simply reinstall the driver.
My computer had booting loop error+ethernet not working + driver crashed after reboot..

from isaacgymenvs.

I met the same situations Segmentation fault, and all the words displayed in the window of Isaac Gym were too vague to distinguish. Have you solved it？

from isaacgymenvs.

I solved the Segmentation fault problem by changing the nvidia-driver to 510.85.02 and cuda version to 11.6.

from isaacgymenvs.

I am currently experiencing this issue! in Isaac sim example https://docs.omniverse.nvidia.com/app_isaacsim/app_isaacsim/tutorial_intro_workflows.html#
./python.sh standalone_examples/api/omni.isaac.franka/follow_target_with_rmpflow.py
I am on ubuntu 22.04 should I install Cuda drivers as well? I just installed 5.25.x.x

from isaacgymenvs.

I have the same issue. After debugging, I found that the issue was actually due to the lack of Vulkan installation. The official Vulkan is more complete on Ubuntu 20.04+, so if there are no special requirements, you can consider using Ubuntu 20.04+. Otherwise, you may need to install Vulkan from source code.

from isaacgymenvs.

I also met this problem when I run the example(p_ython 1080_balls_of_solitude.py_) with remote sever.But I can run the example in my personal computer.The diffrence of environment are the version of cuda and nvidia driver.Why?So confused.

from isaacgymenvs.

I also met this problem when I run the example(p_ython 1080_balls_of_solitude.py_) with remote sever.But I can run the example in my personal computer.The diffrence of environment are the version of cuda and nvidia driver.Why?So confused.

It is possible that the remote server does not have GUI remote services, and when using Isaac, the startup items that depend on GUI training are not closed (but I also haven't understood how to close GUI to to simulation and training).

from isaacgymenvs.

I also met this problem when I run the example(p_ython 1080_balls_of_solitude.py_) with remote sever.But I can run the example in my personal computer.The diffrence of environment are the version of cuda and nvidia driver.Why?So confused.

It is possible that the remote server does not have GUI remote services, and when using Isaac, the startup items that depend on GUI training are not closed (but I also haven't understood how to close GUI to to simulation and training).

Thanks for your answer!But our remote server does have a GUI installed.You mean that when I run example like 1080_balls_of_solitude.py, I must shut down the remote server GUI?

from isaacgymenvs.

I also met this problem when I run the example(p_ython 1080_balls_of_solitude.py_) with remote sever.But I can run the example in my personal computer.The diffrence of environment are the version of cuda and nvidia driver.Why?So confused.

It is possible that the remote server does not have GUI remote services, and when using Isaac, the startup items that depend on GUI training are not closed (but I also haven't understood how to close GUI to to simulation and training).

Thanks for your answer!But our remote server does have a GUI installed.You mean that when I run example like 1080_balls_of_solitude.py, I must shut down the remote server GUI?

emmm,i mean that when you tryrun example like 1080_balls_of_solitude.py,your server must have a GUI

from isaacgymenvs.

I also met this problem when I run the example(p_ython 1080_balls_of_solitude.py_) with remote sever.But I can run the example in my personal computer.The diffrence of environment are the version of cuda and nvidia driver.Why?So confused.

It is possible that the remote server does not have GUI remote services, and when using Isaac, the startup items that depend on GUI training are not closed (but I also haven't understood how to close GUI to to simulation and training).

Thanks for your answer!But our remote server does have a GUI installed.You mean that when I run example like 1080_balls_of_solitude.py, I must shut down the remote server GUI?

because found that the issue was actually due to the lack of Vulkan installation(GUI underlying driver

from isaacgymenvs.

(GUI underlying drive

Thank you, I naively thought ubuntu20.04 came with Vulkan.

from isaacgymenvs.

Here is my experience: I try to train a policy for legged gym and set headless = False to visualize the training process.
When waiting for loading the terrains into isaac gym, it throws segmentation fault (core dumped), after waiting for about 1 minute.
But when I reduce the number of terrains, Isaac Gym load the terrains within 1 minute and it works fine.
In both case, my GPU memory is not full.
So, I guess there is a time limits for loading terrain triangles.

I do not solve this problem.

from isaacgymenvs.