Figure out why nose2 can't run cytoflowgui/tests on Github linux runners

[bpteague]

The error we get is:

Run source activate cytoflow
  source activate cytoflow
  nose2 -v -c package/nose2.cfg -s cytoflowgui/tests -GUI        
  shell: /bin/bash -e {0}
  env:
    pythonLocation: /opt/hostedtoolcache/Python/3.6.12/x64
    LD_LIBRARY_PATH: /opt/hostedtoolcache/Python/3.6.12/x64/lib
    OMP_NUM_THREADS: 1
WARNING:package.nose_setup:Loading customized nose2 configuration
testAddChannel (test_pca.TestPCA) ... libgcc_s.so.1 must be installed for pthread_cancel to work
libgcc_s.so.1 must be installed for pthread_cancel to work
/usr/share/miniconda/envs/cytoflow/lib/python3.6/multiprocessing/semaphore_tracker.py:143: UserWarning: semaphore_tracker: There appear to be 3 leaked semaphores to clean up at shutdown
  len(cache))
/home/runner/work/_temp/9ebd9249-6509-4f9d-b8da-e9823f3d87d3.sh: line 2:  5788 Aborted                 (core dumped) nose2 -v -c package/nose2.cfg -s cytoflowgui/tests -GUI
Error: Process completed with exit code 134.

[kristjaneerik]

source activate cytoflow vs conda activate cytoflow? Or are they the same here?

I guess the pythonLocation and LD_LIBRARY_PATH values shown do not involve the conda environment because that's before the call to source activate cytoflow.

Could add a call to env and also maybe locate libgcc_s.so.1 (may not work though, but something similar using find could) to the list of commands run before nose2:

source activate cytoflow
env
locate libgcc_s.so.1
nose2 -v -c package/nose2.cfg -s cytoflowgui/tests -GUI

Knowing what the true LD_LIBRARY_PATH value is when nose2 is called is helpful; then we'd need to know whether libgcc_s.so.1 is in a path included in there.

Could be a mismatch between e.g. a 32 bit GCC installed being run with a 64 bit python or something like that.

[bpteague]

Tried LD_PRELOAD=/lib/x86_64-linux-gnu/libgcc_s.so.1 but that didn't fix it.

Relevant conversation: https://bugs.python.org/issue44434