Not all PR jobs are getting run on main

[ezyang]

On https://hud.pytorch.org/pr/99469 I have this config run on my job: linux-bionic-cuda11.8-py3.10-gcc7-sm86 / test (default, 2, 5, linux.g5.4xlarge.nvidia.gpu)

However, picking a random commit from main, e.g., https://hud.pytorch.org/pytorch/pytorch/commit/89d57f269fbcd0f7e231ec6ca0c922c97951666d this config does not show up at all (there is a cuda11.8 build, but it is not the same.)

What's going on?

[atalman]

@ezyang linux-bionic-cuda11.8-py3.10-gcc7-sm86 is part of inductor.yml and you have ciflow/inductor label on #99469 , however on #102993 no ciflow/inductor label attached, hence the above test was not executed

[atalman]

Looks like linux-bionic-py3.8-clang9, linux-bionic-py3.11-clang9, linux-bionic-cuda11.8-py3.10-gcc7-sm86 tests on #102640 are successful here: https://hud.pytorch.org/pr/102640

[ezyang]

Are you telling me that ciflow/inductor jobs NEVER run on main? That seems bad!

[huydhn]

For this case, the job on trunk commit has been migrated to CUDA 12.1 by pytorch/pytorch#102178. So its name is now linux-bionic-cuda12.1-py3.10-gcc9-sm86 / test (default, 2, 5, linux.g5.4xlarge.nvidia.gpu). You could do a rebase on your PR, then 11.8 job will be replaced by 12.1 one