Import mxnet throws libcudart error on non cuda linux hosts

[vdantu]

I am seeing

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/mxnet-model-server/testenv/local/lib/python2.7/site-packages/mxnet/__init__.py", line 25, in <module>
    from . import engine
  File "/mxnet-model-server/testenv/local/lib/python2.7/site-packages/mxnet/engine.py", line 23, in <module>
    from .base import _LIB, check_call
  File "/mxnet-model-server/testenv/local/lib/python2.7/site-packages/mxnet/base.py", line 113, in <module>
    _LIB = _load_lib()
  File "/mxnet-model-server/testenv/local/lib/python2.7/site-packages/mxnet/base.py", line 105, in _load_lib
    lib = ctypes.CDLL(lib_path[0], ctypes.RTLD_LOCAL)
  File "/usr/lib/python2.7/ctypes/__init__.py", line 362, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: libcudart.so.9.0: cannot open shared object file: No such file or directory

When I try to import mxnet after doing a pip install from setup.py. This happens when I try to import "mxnet" in my python module. This doens't happen with mxnet-mkl and happens with mxnet-cu90mkl package on generic linux host. We might need to change the default installation for linux.

Steps to recreate this issue on a linux machine:

1. (testenv)$ pip install mxnet-cu90mkl
2. (testenv)$ python
3.  >>> import mxnet

[vrakesh]

This issue, seems to appearing in CI as well, My recent pull request is failing in codebuild (linux base) due to the very same reason, any changes to the cu90mkl package recently??, the change was tested on many linux machines before, this seems to be something new. Will look into it

[vrakesh]

resolved in PR #394

[nithishc829]

Solution :

1. deactivate the environment
2. edit the bashrc file like nano ~/.bashrc file & add the cuda path to environment variable LD_LIBRARY_PATH LD_LIBRARY_PATH=/usr/local/cuda-9.0/lib64:$LD_LIBRARY_PATH and save
3. type the command source ~/.bashrc
4. type the command sudo ldconfig to reflect the changes for library to be seen by OS
5. done! activate the environment & run python
6. import mxnet
   Problem solved. I spent 1 whole day find the solution...