Environment builds but some packages have libstdc++.so.6 errors about cookiecutter-cms

[slochower]

Yep, thanks.

[loriab]

Possibly need the same compiler toolchain for building the local package as was used to build the env packages. Another suspect is that the c-f compiler toolchain has switched over the past months, so old envs may behave differently than new envs.

[Lnaden]

I might have an idea, but will need some more information. Can you post your test_env.yaml file?

I will point out that the cookiecutter does not do anything with compilation, so this might be an upstream package issue, not necessarily the cookiecutter itself. For instance, the line /home/davids4/anaconda3/envs/paprika-debug-tleap-dummy/lib/python3.6/site-packages/scipy/sparse/_sparsetools.cpython-36m-x86_64-linux-gnu.so indicates it might be an issue with the scipy download, which seems unlikely.

Are there additional packages in your base/root conda environment which may get overwritten or temporarily removed when you activate the new environment?

[slochower]

Hi Levi, thanks for your thoughts. Here is the test_env.yaml file:

(paprika-debug-tleap-dummy) [davids4@tscc-login1 pAPRika]$ cat devtools/conda-envs/test_env.yaml 
name: paprika-debug-tleap-dummy
channels:
  - conda-forge
  - omnia
  # AmberTools seem to live at this repository...
  - http://ambermd.org/downloads/ambertools/conda/
dependencies:
    # Base depends
  - python=3.6
  - pip
  - numpy
  - scipy
  - parmed
  - mdtraj
  - pymbar
  - openmm
  # Eventually we can/should migrate away from this, once AMBER gets onto omnia/conda-forge.
  # See https://openforcefieldgroup.slack.com/archives/C8NE3J96U/p1549403299386500
  - ambertools=18

    # Testing
  - pytest
  - pytest-cov

    # Pip-only installs
  - pip:
    - codecov

I'm not sure if there's anything in base that might be overwritten by this test environment. It's been a few months since I've worked on TSCC with this code, so I just did a conda update conda and conda update before trying to test my cookiecutter branch. I'm running conda 4.6.3 now. FWIW, here's the output of conda list in my test environment:

`conda list`

(paprika-debug-tleap-dummy) [davids4@tscc-login1 pAPRika]$ conda list
# packages in environment at /home/davids4/anaconda3/envs/paprika-debug-tleap-dummy:
#
# Name                    Version                   Build  Channel
ambertools                18.0                          0    omnia
atomicwrites              1.3.0                      py_0    conda-forge
attrs                     18.2.0                     py_0    conda-forge
blas                      1.1                    openblas    conda-forge
blosc                     1.15.1            hf484d3e_1002    conda-forge
bzip2                     1.0.6             h14c3975_1002    conda-forge
ca-certificates           2018.11.29           ha4d7672_0    conda-forge
certifi                   2018.11.29            py36_1000    conda-forge
chardet                   3.0.4                    pypi_0    pypi
codecov                   2.0.15                   pypi_0    pypi
coverage                  4.5.2           py36h14c3975_1000    conda-forge
fftw3f                    3.3.4                         2    omnia
hdf5                      1.10.4          nompi_h11e915b_1105    conda-forge
idna                      2.8                      pypi_0    pypi
libffi                    3.2.1             hf484d3e_1005    conda-forge
libgcc                    7.2.0                h69d50b8_2    conda-forge
libgcc-ng                 7.3.0                hdf63c60_0    conda-forge
libgfortran-ng            7.2.0                hdf63c60_3    conda-forge
libstdcxx-ng              7.3.0                hdf63c60_0    conda-forge
lzo                       2.10              h14c3975_1000    conda-forge
mdtraj                    1.9.2           py36hff88cb2_1000    conda-forge
mmpbsa-py                 16.0                     pypi_0    pypi
more-itertools            4.3.0                 py36_1000    conda-forge
ncurses                   6.1               hf484d3e_1002    conda-forge
numexpr                   2.6.9           py36h637b7d7_1000    conda-forge
numpy                     1.16.1          py36_blas_openblash1522bff_0  [blas_openblas]  conda-forge
openblas                  0.3.3             h9ac9557_1001    conda-forge
openmm                    7.3.0           py36_cuda92_rc_1    omnia
openssl                   1.0.2p            h14c3975_1002    conda-forge
packmol-memgen            0.9.9a0                  pypi_0    pypi
pandas                    0.24.1           py36hf484d3e_0    conda-forge
paprika                   0.0.3+87.g1ecf2fe.dirty           dev_0    <develop>
parmed                    3.0.0+57.g74a84d30          pypi_0    pypi
pdb4amber                 1.6.dev0                 pypi_0    pypi
pip                       19.0.2                   py36_0    conda-forge
pluggy                    0.8.1                      py_0    conda-forge
py                        1.7.0                      py_0    conda-forge
pymbar                    3.0.3            py36h3010b51_3    conda-forge
pymsmt                    18.0                     pypi_0    pypi
pytables                  3.4.4           py36h5d69a80_1004    conda-forge
pytest                    4.2.1                    py36_0    conda-forge
pytest-cov                2.6.1                      py_0    conda-forge
python                    3.6.7             hd21baee_1001    conda-forge
python-dateutil           2.8.0                      py_0    conda-forge
pytraj                    2.0.3                    pypi_0    pypi
pytz                      2018.9                     py_0    conda-forge
readline                  7.0               hf8c457e_1001    conda-forge
requests                  2.21.0                   pypi_0    pypi
sander                    16.0                     pypi_0    pypi
scipy                     1.2.0           py36_blas_openblash1522bff_1201  [blas_openblas]  conda-forge
setuptools                40.8.0                   py36_0    conda-forge
six                       1.12.0                py36_1000    conda-forge
sqlite                    3.26.0            h67949de_1000    conda-forge
tk                        8.6.9             h84994c4_1000    conda-forge
urllib3                   1.24.1                   pypi_0    pypi
wheel                     0.33.0                   py36_0    conda-forge
xz                        5.2.4             h14c3975_1001    conda-forge
zlib                      1.2.11            h14c3975_1004    conda-forge

[Lnaden]

I think this might actually be an issue with some pointer to old linkers which the scipy library is looking at, but expecting a newer version which should get shipped with Conda.

I don't have a good answer for you, but here is a link to someone having this exact same issue: ContinuumIO/anaconda-issues#5191 (comment)

From what I am reading on that issue, it looks like it might be your LD_LIBRARY_PATH variable is set to point at some old libgcc files and so the SciPy binaries are looking there first instead of at the ones installed in the Conda Environment.

[slochower]

Hmm, yes indeed, good sleuthing. I didn't have anything overriding LD_LIBRARY_PATH in my .bashrc but it is set by the cluster and some of those look kinda old. If I manually add the conda libraries to the front of that, things do work. For example:

(paprika-debug-tleap-dummy) [davids4@tscc-login1 lib]$ LD_LIBRARY_PATH=/home/davids4/anaconda3/lib:$LD_LIBRARY_PATH

(I guess I could have used the environment specific lib directory, but the base one worked). It's not ideal, but it's a viable workaround.

[Lnaden]

Glad to hear. Can we close this issue then as it seems to be beyond the scope of Cookiecutter?