Migrate TPU tests to GitHub actions (#14687)

* Migrate TPU tests to GitHub actions

* No working dir

* Keep _target

* Dont skip draft

* CHECK_SLEEP

* Not yet

* Remove recurrent cleanup script

* Set secrets

* a step cannot have both the `uses` and `run` keys

* Version $PYTHON_VER was not found in the local cache

* can't load package ... ($GOPATH not set)

* The `set-env` command is disabled

* Try updating go

* Match timeout

* simplify path

* More cleanup

* Install coverage. Unmark draft

* Update .github/workflows/ci-pytorch-test-tpu.yml

* DEBUG echo

* Revert "DEBUG echo"

This reverts commit 4011856.

* More debug

* SSH

* Im stupid

* Remove always()

* Forgot some

Co-authored-by: Jirka Borovec <[email protected]>
Co-authored-by: Luca Antiga <[email protected]>

.github/CODEOWNERS

@@ -9,7 +9,6 @@
 
 # CI/CD and configs
 /.github/       @borda @carmocca @akihironitta @otaj
-/.circleci/     @borda @carmocca @akihironitta @otaj
 /.azure/        @borda @carmocca @akihironitta @otaj
 /dockers/       @borda @carmocca @akihironitta @otaj
 *.yml           @borda @carmocca @akihironitta @otaj

.github/checkgroup.yml

@@ -6,13 +6,6 @@ subprojects:
 
   # SECTION: pytorch_lightning
 
-  - id: "CI: CircleCI"
-    paths:
-      - ".circleci/**"
-      - ".github/workflows/ci-circleci.yml"
-    checks:
-      - "test-on-tpus"
-
   - id: "pytorch_lightning"
     paths:
       # all examples don't need to be added because they aren't used in CI, but these are
@@ -58,7 +51,7 @@ subprojects:
       - "pl-slow (macOS-11, 3.7, 1.11)"
       - "pl-slow (ubuntu-20.04, 3.7, 1.11)"
       - "pl-slow (windows-2022, 3.7, 1.11)"
-      # TODO: since this job cannot run on forks, it cannot be required or it will block all PL PRs from forks
+      # TODO: since this job has intermittent availability, it cannot be required or it will block all PL PRs from forks
       #- "test-on-tpus"
 
   - id: "pytorch_lightning: CPU"
@@ -110,6 +103,12 @@ subprojects:
     checks:
       - "pytorch-lightning (IPUs)"
 
+  - id: "pytorch-lightning: TPU"
+    paths:
+      - ".github/workflows/tpu-tests.yml"
+    checks:
+      - "test-on-tpus"
+
   - id: "pytorch_lightning: Docs"
     paths:
       - "docs/source-pytorch/**"

.github/workflows/README.md

@@ -12,7 +12,7 @@
 | pytorch-lightning (HPUs)   | .azure-pipelines/hpu-tests.yml             | Run only HPU-specific tests.                                                                                                                                                | HPU           |
 | pytorch-lightning (GPUs)   | .azure-pipelines/gpu-tests-pytorch.yml     | Run all CPU and GPU-specific tests, standalone, and examples. Each standalone test needs to be run in separate processes to avoid unwanted interactions between test cases. | GPU           |
 | PyTorchLightning.Benchmark | .azure-pipelines/gpu-benchmark.yml         | Run speed/memory benchmarks for parity with pure PyTorch.                                                                                                                   | GPU           |
-| test-on-tpus               | .circleci/config.yml                       | Run only TPU-specific tests.                                                                                                                                                | TPU           |
+| test-on-tpus               | .github/workflows/tpu-tests.yml            | Run only TPU-specific tests.                                                                                                                                                | TPU           |
 
 - \*Accelerators used in CI
 

.github/workflows/ci-pytorch-dockers.yml

@@ -56,7 +56,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        # the config used in '.circleci/config.yml`'
+        # the config used in '.github/workflows/tpu-tests.yml'
         python_version: ["3.7"]
         xla_version: ["1.12"]
     steps: