Github CI pytest network-dependent error

[Sevans711]

Version

2026.6.0

How did you install UXarray?

Source

What happened?

CI pipelines failing recently due to network https issues, but only 1 out of 9 runs failed. Examples:

• https://github.com/UXARRAY/uxarray/actions/runs/29851989004 from Bump the actions group with 2 updates #1601. Here, Python 3.12 ubuntu failed, but all others tests passed.
  • Error message: FAILED test/io/test_structured.py::test_read_structured_grid_from_ds[ersstv5] - requests.exceptions.ReadTimeout: HTTPSConnectionPool(host='raw.githubusercontent.com', port=443): Read timed out. (read timeout=30)
• https://github.com/UXARRAY/uxarray/actions/runs/29864267121 from Reduce imports inside functions #1603. Here, Python 3.13 macos failed, but all other tests passed.
  • Error message: FAILED test/io/test_structured.py::test_read_structured_grid_from_ds[ersstv5] - requests.exceptions.HTTPError: 503 Server Error: first byte timeout for url: https://raw.githubusercontent.com/pydata/xarray-data/master/ersstv5.nc

Raising this issue primarily because it occurred twice recently, and it's not trivial to find the "rerun pipeline for this PR" button.

What did you expect to happen?

I expected CI pipeline to not be as finicky with this dependence on network connection. Maybe it should retry test automatically if the failure was just an https network issue?

Can you provide a MCVE to repoduce the bug?

[Sevans711]

For future reference, here is where to find the "rerun pipeline for this PR" button. Note: it might be necessary to wait for all pending jobs to complete, before this button becomes available.

Feel free to mark this issue as closed if the intention is to accept that tests might be finicky, and developers should be expected to find and push this button when the failure is due to a random network error. Although, retrying once automatically on network-related failure might still make sense to implement? Happy to let others decide, now that I know where the "rerun pipeline for this PR" button lives.

First, click the failed job:

Next, scroll all the way to the top of the page, then use the "Re-run failed jobs" button at the top right: