cpu (macOS-10.15, 3.7, latest, stable) gets cancelled due to timeout

[akihironitta]

🐛 Bug

The CI run always (?) fails with the following message due to timeout (set to 40 min). Similarly to #11184, the actual tests seem to always pass, but the difference is that runs here get cancelled due to timeouts instead of segfaults.

=== 2247 passed, 499 skipped, 11 xfailed, 3618 warnings in 613.75s (0:10:13) ===
Error: The operation was canceled.

To Reproduce

• https://github.com/PyTorchLightning/pytorch-lightning/runs/4601898496?check_suite_focus=true#step:16:2860
• https://github.com/PyTorchLightning/pytorch-lightning/runs/4601604324?check_suite_focus=true#step:16:2860
• https://github.com/PyTorchLightning/pytorch-lightning/runs/4601145564?check_suite_focus=true#step:16:2860

Expected behavior

No timeout.

Environment

• PyTorch Lightning Version (e.g., 1.5.0):
• PyTorch Version (e.g., 1.10):
• Python version (e.g., 3.9):
• OS (e.g., Linux):
• CUDA/cuDNN version:
• GPU models and configuration:
• How you installed PyTorch (conda, pip, source):
• If compiling from source, the output of torch.__config__.show():
• Any other relevant information:

Additional context

Could be the same issue as #11184, but creating this issue for tracking just in case.

cc @carmocca @akihironitta @Borda

[carmocca]

Looking at the time difference from the timestamps:

Wed, 22 Dec 2021 04:31:25 GMT === 2239 passed, 509 skipped, 11 xfailed, 3589 warnings in 531.92s (0:08:51) ===
Wed, 22 Dec 2021 04:54:05 GMT Error: The operation was canceled.

Got stuck for 23 minutes after finishing the tests.
It's likely linked to #11184

[akihironitta]

Closing this since the issue's been resolved in master somehow.