Scheduled workflow failed

[github-actions]

Oh no, something went wrong in the scheduled workflow PyTorch version tests with commit 43f94a8.
Please look into it:

https://github.com/pytorch/ignite/actions/runs/1613520299

Feel free to close this if this was just a one-off error.

[vfdev-5]

@bibhabasumohapatra looks like there is an issue with a test for pytorch 1.8.1 and 1.7.1 with python 3.9
For example for 1.8.1

>       outputs = F.conv2d(input_list, self._kernel, groups=channel)
E       RuntimeError: std::bad_alloc

Can you check if it was a pytorch bug that is fixed on 1.10.X or there is something also from our side ?

[bibhabasumohapatra]

Okay, looking into it!

[bibhabasumohapatra]

again /// ig only problem is 3.9 and 1.7.1

/home/runner/work/_temp/1b5abbbf-514e-4a96-8012-d3ecd4833c4a.sh: line 1:  2814 Segmentation fault      (core dumped) pytest tests -vvv
Error: Process completed with exit code 139.

I ran again ... with pulled recent commits!

I don't know but in my forked one its not failing : https://github.com/bibhabasumohapatra/ignite/actions/runs/1620567909

[vfdev-5]

Link found by @bibhabasumohapatra : pytorch/pytorch#70237
Maybe related to the failures for 3.9, 1.8.1

[bibhabasumohapatra]

Link found by @bibhabasumohapatra : pytorch/pytorch#70237 Maybe related to the failures for 3.9, 1.8.1

#2387
in this there is no more 3.9 ,1.8.1 failure, it doesn't come up now!

[vfdev-5]

@bibhabasumohapatra can you disable the tests for 3.9 ,1.8.1 and 3.9 ,1.7.1 and mention this issue in the comment. Thanks.
Otherwise, there are too much notifications every day.