All NotebooksTestCase tests fail with CalledProcessError

[jneuff]

Relevant output:

CalledProcessError: Command '['jupyter', 'nbconvert', '--to', 'notebook', '--execute', '--ExecutePreprocessor.timeout=600', '--ExecutePreprocessor.kernel_name=python2', '--output', '/tmp/tmpu50Tiu.ipynb', 'notebooks/human_activity_recognition_multi_class_example.ipynb']' returned non-zero exit status 1

/usr/lib64/python2.7/subprocess.py:186: CalledProcessError
----------------------------------------- Captured stderr call ------------------------------------------
[NbConvertApp] Converting notebook notebooks/human_activity_recognition_multi_class_example.ipynb to notebook
[NbConvertApp] Executing notebook with kernel: python2
[NbConvertApp] WARNING | Native kernel (python2) is not available

and

File "/home/user/.venvs/tsf2/lib/python2.7/site-packages/jupyter_client/kernelspec.py", line 203, in get_kernel_spec
    raise NoSuchKernel(kernel_name)
jupyter_client.kernelspec.NoSuchKernel: No such kernel named python2

I have the same problem with running the tests with Python 3.
The relevant code of the tests seems to be:

        if six.PY2:
            args += ["--ExecutePreprocessor.kernel_name=python2"]
        elif six.PY3:
            args += ["--ExecutePreprocessor.kernel_name=python3"]

Apparently these settings don't work on my machine (Fedora 26).

[MaxBenChrist]

Is ipython installed?

I have seen that error when I have only installed jupyter

[nikhase]

I've had a similar issue. Notebook tests failed on my local machine. The main line of interest in my case was jupyter_client.kernelspec.NoSuchKernel: No such kernel named py27tsfresh.

I am working with Conda and it's environments; the same issue was discussed here. I adapted the solution, and it worked!

In my case, the system's python installation python2 and python3 somehow "masked" my properly configured environment with tsfresh. I did

$ jupyter kernelspec list
Available kernels:
  python2    /home/niklas/.local/share/jupyter/kernels/python2
  python3    /home/niklas/.local/share/jupyter/kernels/python3

These are links to the system's python installations (without tsfresh and dependencies installed), which were apparently called by nbconvert in test_notebooks.py. You might see other kernels as well if you configured to use R, Julia or whatever language in your notebooks. I think, these are the ones visible in the Dropdown-Menu you can choose your kernels from when you open a notebook. However, after

$ jupyter kernelspec remove python2
$ jupyter kernelspec remove python3

I ran the test python setup.py test, and it worked! Now it looks like

$ jupyter kernelspec list
Available kernels:
  python2    /home/niklas/anaconda3/envs/py27tsfresh/lib/python2.7/site-packages/ipykernel/resources

I don't know if it's the same in virtualenv. Maybe you can list your kernels and see if it is somewhere in /home/user/.venvs/tsf2/lib/python2.7/site-packages/....

Use it on your own risk. Please let me now what the issue was.

This solves my post / experience I reported in a former issue as well.

[nikhase]

@MaxBenChrist do you run tests locally? Do they pass? What is your setup? virtualenv, system python or conda? Thx for info.

[MaxBenChrist]

@nikhaas virtualenv, Mac OS or Ubuntu 16.04, no problems so far.

[jneuff]

Thanks for the hints @nikhaas and @MaxBenChrist. However, the strange thing is, that running the tests at a later time actually worked out. I do not have the time to investigate this, but we should keep in mind, that this can be a source of errors.

[MaxBenChrist]

As this error did not show up again and we excluded notebook tests from the test suite I close this now