Getting kill signal during Actions CI


I am getting a weird kill signal error when performing CI for this PR with the following error [ref]:

  pytest --cov=. --cov-report=xml
  shell: /usr/bin/bash -e {0}
    pythonLocation: /opt/hostedtoolcache/Python/3.7.12/x64
    LD_LIBRARY_PATH: /opt/hostedtoolcache/Python/3.7.12/x64/lib
============================= test session starts ==============================
platform linux -- Python 3.7.12, pytest-7.0.1, pluggy-1.0.0
rootdir: /home/runner/work/GaNDLF/GaNDLF
plugins: cov-3.0.0
collected 30 items

/home/runner/work/_temp/ line 1:  2930 Killed                  pytest --cov=. --cov-report=xml
testing/ .......
Error: Process completed with exit code 137.

I have also tried updating setup-python to v3.0.0 but I am getting the same error.

Any help would be greatly appreciated!


Does it work locally?

Yes, it works locally without issue.

Not sure but I did notice two things, no idea if they’re relevant or not.

  1. The failed job seems to use python 3.7, while the black job seems to use 3.8
  2. The part you added to full_test refers to efficientnetb0, lowercase b elsewhere it seems uppercase.

Edit: oh #2 might just be a matter of import vs use :slightly_frowning_face:

To answer your points:

  1. Hasn’t been an issue with any other PR (either before or after this particular PR)
  2. We define lowercase dictionary keys for all networks, so it gets handled there [ref]

If those would have been the issues, it would have failed while the tests would have been running (see the PRs that passed). Here, the tests don’t seem to begin.

EDIT: PRs that passed both before and after this problematic PR:

1 Like

I think I figured out the issue. It is a memory issue because the runners have 2G of RAM available. Is there any way to increase this to 3 or perhaps 4G for the Linux runners?

1 Like

Only if you setup your own runner. Are you able to break the tests up so they use less memory?

That’s what I am going to try to do, thanks for your help!