Merge branch 'main' into 2023-11-29-meeting-notes

Author: jakirkham

.github/workflows/deploy.yml

@@ -14,7 +14,7 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - uses: conda-incubator/setup-miniconda@v2
+      - uses: conda-incubator/setup-miniconda@v3
         with:
           python-version: 3.8
           channels: conda-forge

src/maintainer/knowledge_base.rst

@@ -1668,6 +1668,32 @@ The steps involved are, roughly:
    consider merging the PR opened at step 2 now so it can apply to all the downstream feedstocks.
 
 
+.. _opengpuserver:
+
+Packages that require a GPU or long-running builds 
+==================================================
+
+conda-forge has access to `an OpenStack server <https://github.com/Quansight/open-gpu-server>`__ that provides GPU builds and long-running builds (beyond the usual 6h limit).
+If your package needs a GPU to be built or tested, or its compilation times are so long that they are currently done manually off-CI, you can request access to these runners.
+To do so:
+
+1. Open a PR in `conda-forge/admin-requests <https://github.com/conda-forge/admin-requests>`__. Follow the instructions in the repository README. 
+   Note you need to request the type of resource you want access to (e.g. GPU runners, or long-running CPU builds)
+   Once merged, this will enable the self-hosted Github Actions runners for your feedstock.
+2. In order to trigger jobs for these runners, the maintainer must have read and agreed to the open-gpu-server `terms of use <https://github.com/Quansight/open-gpu-server/blob/main/TOS.md>`__.
+   You will need to open a PR in the open-gpu-server repository, as instructed in their README. 
+   You only need to do this once per maintainer (e.g. if you maintain multiple feedstocks).
+3. Finally, you can configure your feedstock to use the self-hosted runners. A PR will have been created by admin-requests after the PR in step (1) is merged.
+   However, due to security measurements imposed by Github, automated re-rendering is not possible when they modify Github Actions workflows.
+   You will need to rerender it manually by running ``conda-smithy rerender`` in your machine and then commit and push the result.
+
+.. note::
+
+  Due to some technical and legal limitations, some of the usual automation infrastructure is not available for these runners.
+  As mentioned above, the conda-forge bots won't be able to rerender your feedstock automatically anymore.
+  Automerge will not function properly either. Also note that the conda-forge bots won't be able to trigger the self-hosted runners. 
+  Closing and reopening the PR won't work, but a maintainer with sufficient permissions can trigger it manually by pushing an empty commit. 
+
 .. _osxarm64:
 
 Apple Silicon builds
@@ -1833,12 +1859,11 @@ higher than the package uploaded with the ``dev`` label.
 
 How to update your feedstock token?
 ====================================
-To reset your feedstock token and fix issues with uploads, follow these steps:
 
-1. Create a new text file in the ``token_reset`` directory of the ``conda-forge/admin-requests`` repo.
-2. Add the name of your feedstock in the text file. While adding the name, don't add "-feedstock" to the end of it. For example: for ``python-feedstock``, just add ``python``.
+To reset your feedstock token and fix issues with uploads, follow these steps:
 
-See `token_reset/example.txt <https://github.com/conda-forge/admin-requests/blob/main/token_reset/example.txt>`__ for an example.
+1. Go to the ``conda-forge/admin-requests`` repo and copy `examples/example-broken.yml <https://github.com/conda-forge/admin-requests/blob/main/examples/example-broken.yml>`__ to the ``requests/`` folder.
+2. Add the name of your feedstock in the YML file. While adding the name, don't add "-feedstock" to the end of it. For example: for ``python-feedstock``, just add ``python``.
 
 .. _using_arch_rebuild:
 

src/maintainer/updating_pkgs.rst

@@ -255,8 +255,8 @@ If instead the actual contents of the package are broken, the following steps wi
 remove broken packages from the ``main`` channel:
 
 1. Locate the paths to broken files on `anaconda.org <https://anaconda.org>`__, by searching for the conda-forge package and switching to the files tab.
-2. Fork `conda-forge/admin-requests <https://github.com/conda-forge/admin-requests>`__ and add a new text file in the ``broken`` directory.
-3. Add the broken files to the new text file, one path per line. See `broken/example.txt <https://github.com/conda-forge/admin-requests/blob/main/broken/example.txt>`__ for an example file.
+2. Fork `conda-forge/admin-requests <https://github.com/conda-forge/admin-requests>`__ and add a new YML file in the ``requests`` directory.
+3. Add the broken files to the new YML document.See `examples/example-broken.yml <https://github.com/conda-forge/admin-requests/blob/main/examples/example-broken.yml>`__ for an example file.
 4. Open a new PR. Once merged, a bot will label all listed files as broken, thus effectively removing them from the channel.