README.md
1# pytorch/.github
2
3> NOTE: This README contains information for the `.github` directory but cannot be located there because it will overwrite the
4repo README.
5
6This directory contains workflows and scripts to support our CI infrastructure that runs on GitHub Actions.
7
8## Workflows
9
10- Pull CI (`pull.yml`) is run on PRs and on main.
11- Trunk CI (`trunk.yml`) is run on trunk to validate incoming commits. Trunk jobs are usually more expensive to run so we do not run them on PRs unless specified.
12- Scheduled CI (`periodic.yml`) is a subset of trunk CI that is run every few hours on main.
13- Binary CI is run to package binaries for distribution for all platforms.
14
15## Templates
16
17Templates written in [Jinja](https://jinja.palletsprojects.com/en/3.0.x/) are located in the `.github/templates` directory
18and used to generate workflow files for binary jobs found in the `.github/workflows/` directory. These are also a
19couple of utility templates used to discern common utilities that can be used amongst different templates.
20
21### (Re)Generating workflow files
22
23You will need `jinja2` in order to regenerate the workflow files which can be installed using:
24```bash
25pip install -r .github/requirements/regenerate-requirements.txt
26```
27
28Workflows can be generated / regenerated using the following command:
29```bash
30.github/regenerate.sh
31```
32
33### Adding a new generated binary workflow
34
35New generated binary workflows can be added in the `.github/scripts/generate_ci_workflows.py` script. You can reference
36examples from that script in order to add the workflow to the stream that is relevant to what you particularly
37care about.
38
39Different parameters can be used to achieve different goals, i.e. running jobs on a cron, running only on trunk, etc.
40
41#### ciflow (trunk)
42
43The label `ciflow/trunk` can be used to run `trunk` only workflows. This is especially useful if trying to re-land a PR that was
44reverted for failing a `non-default` workflow.
45
46## Infra
47
48Currently most of our self hosted runners are hosted on AWS, for a comprehensive list of available runner types you
49can reference `.github/scale-config.yml`.
50
51Exceptions to AWS for self hosted:
52* ROCM runners
53
54### Adding new runner types
55
56New runner types can be added by committing changes to `.github/scale-config.yml`. Example: https://github.com/pytorch/pytorch/pull/70474
57
58> NOTE: New runner types can only be used once the changes to `.github/scale-config.yml` have made their way into the default branch
59
60### Testing [pytorch/builder](https://github.com/pytorch/builder) changes
61
62In order to test changes to the builder scripts:
63
641. Specify your builder PR's branch and repo as `builder_repo` and `builder_branch` in [`.github/templates/common.yml.j2`](https://github.com/pytorch/pytorch/blob/32356aaee6a77e0ae424435a7e9da3d99e7a4ca5/.github/templates/common.yml.j2#LL10C26-L10C32).
652. Regenerate workflow files with `.github/regenerate.sh` (see above).
663. Submit fake PR to PyTorch. If changing binaries build, add an appropriate label like `ciflow/binaries` to trigger the builds.
67