test_thread_limit_apply FAILED assert statement

[nruloff]

I'm installing MDAnalysis for the first time. The installation is in a Linux environment. Running the code block from the Quick start guide outputs that MDAnalysis is installed and able to successfully run a Universe with 3341 atoms.

I have proceeded to the 'testing' portion of the installation guide, but I am getting a FAILED assert statement from test_base.py

The code of test_base.py looks like it is performing a transformation to set the max number of threads to 2 prior to the assert statement test by "new_trans = CustomTransformation(max_threads=2).

But the thread info seems to be getting its thread count from the default_thread_info / new_trans is not being impacted by the 'max_threads' input of 'CustomTransformation(max_threads=2).

As I am writing this discussion thread - I am also noticing that the 'testing' portion that I am reading from is a subsection of the installation for 'Development Versions'.

• Does this testing check not apply to my installation as I have not installed a 'Development Version'?
• Does the check from running the code block from the Quick start guide suffice for a successful installation?

Thanks for any help in advance!

[RMeli]

Hi @nruloff, it's good you are trying to run all the tests. In principle they should all pass. Looks like you bumped into a new issue: #4570

Do you have more informations about your system and environment (specifically, the NumPy version)?

[IAlibay]

I'm very roughly remembering but I think this might be related to: #4570

Could you confirm what version of threadpoolctl you are using @nruloff ?

[nruloff]

@IAlibay - certainly. The threadpoolctl version that I am using is 2.2.0 - see the screenshot below.

[nruloff]

I should add - the Basic Example test did not show any errors. But I am just starting to read through the Documentation / Tutorials now. If there are any tutorials which specifically test the thread limit, I can test that as well.

[IAlibay]

Thanks! That's a very old version of threadpoolctl, ~ 3 years old. I suspect you might need to go for a newer version, maybe trying a new environment with threadpoolctl ~ 3.3+ could fix the issue.

[nruloff]

@IAlibay - I updated threadpoolctl to 3.5.0 and reran the pytest and still obtain the same error (screenshot below).

I am noticing that when I run the pytest that there are "7 skipped" - could this potentially be related or is this a separate issue?