How to add hard link on the libraries present inside the docker image?

  cuda, docker, python, tensorflow

I am trying to run my model training on GPU server but i am getting below error saying libcusolver.so.11 is not present.

Error: tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library ‘libcusolver.so.11’; dlerror: libcusolver.so.11: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /azureml-envs/azureml_0223502b9e9fa00eac9eae53b2b1cdcd/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64

Looking into the various resolution posted on the below link we need to provide hard link to the libcusolver.so.10.

https://github.com/tensorflow/tensorflow/issues/43947

Issue: I need help in how to provide hard link inside by docker container image.

Thanks for your help in advance.

Source: Docker Questions

LEAVE A COMMENT