LD_LIBRARY_PATH is not respected

Environment data

• VS Code version: 1.51.1

• Jupyter Extension version (available under the Extensions sidebar): !ldd /ccs/home/hzfmer/file_back/programs/andes/anaconda3/lib/python3.8/site-packages/osgeo/_gdal.cpython-38-x86_64-linux-gnu.so | grep libnsl

• Python Extension version (available under the Extensions sidebar): !ldd /ccs/home/hzfmer/file_back/programs/andes/anaconda3/lib/python3.8/site-packages/osgeo/_gdal.cpython-38-x86_64-linux-gnu.so | grep libnsl

• OS (Windows | Mac | Linux distro) and version: Linux-Red Hat 7.6

• Python and/or Anaconda version: 3.8.3-Anaconda

• Type of virtual environment used (N/A | venv | virtualenv | conda | …): conda

• Jupyter server running: Remote

Expected behaviour

I was trying to import a package “gdal” from “osgeo”, which is successful if I ran it in the system or VSCode terminal, while in vscode-jupyter it was unable to find the dynamic library “libnsl.so.1”. It’s strange that ldd in the terminal finds it, but Jupyter doesn’t.

I have added the location of this library to LD_LIBRARY_PATH, and Jupyter seems to read the variable, but somehow doesn’t respect it.

Actual behaviour

Steps to reproduce:

This might be not very likely to reproduce, since I am working on an HPC cluster. I would expect some setting/configuration as a workaround.

[NOTE: Self-contained, minimal reproducing code samples are extremely helpful and will expedite addressing your issue]

1. XXX

Logs

Output for Jupyter in the Output panel (View→Output, change the drop-down the upper-right of the Output panel to Jupyter)

XXX

[hzfmer]

For any one reads this in similar situation, Jupyter via Remote-SSH uses the default SHELL, and even if you specify a different SHELL through RemoteCommand in ssh config, it’s not respected. Please read the comment above to change your environment variable is needed, or change the default shell instead.

[IanMatthewHuff]

Thanks for looping back with the solution in case any one else hits this.