[BUG]: 24.12 nightly wheel tests failing with SSL certificate issues

[jameslamb]

Version

24.12

Which installation method(s) does this occur on?

No response

Describe the bug.

Many nightly wheel tests are failing on 24.12. I see dozens of failures, in two broad categories.

Stuff like this:

FAILED python/cugraph/cugraph/tests/traversal/test_sssp.py::test_sssp_data_type_conversion[1-graph_file0] - KeyError: "results for {'algo': 'single_source_dijkstra_path_length', 'graph_dataset': 'karate', 'graph_directed': 'True', 'source': '1', 'test': 'data_type_conversion', 'resultset_name': 'traversal'} not found"

And stuff like this:

FAILED python/cugraph/cugraph/tests/utils/test_dataset.py::test_download[dataset0] - urllib.error.URLError: <urlopen error [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate (_ssl.c:1007)>

(build link)

These failures only occur on the CUDA 11.8.0, Python 3.10 wheels test jobs.

Minimum reproducible example

See CI links above.

Relevant log output

See CI links above.

Environment details

See CI links above.

Other/Misc.

Because they looked network-related, I did try manually re-running these failed jobs... they failed in exactly the same way.

I know that cugraph tests do some downloading of externally-hosted datasets... maybe one of those sites hosting that data had an outage or changed some configuration?

Code of Conduct

• I agree to follow cuGraph's Code of Conduct
• I have searched the open bugs and have found no duplicates for this bug report

[rlratzel]

I'm wondering if the solution here is relevant for these SSL errors.

[nv-rliu]

Can this be closed now?

[jameslamb]

I think so, yes.

The latest 24.12 test run did fail: https://github.com/rapidsai/cugraph/actions/runs/12349459995

As did the latest 25.02 test run: https://github.com/rapidsai/cugraph/actions/runs/12372518014

Neither looks like this exact issue though. Let's close it. You may want to watch rapidsai/ci-imgs#218. Once it goes out, it might be possible to rever the changes from #4825 or some subset of them.