TorchX-MCAD support (#679)

* TorchX-MCAD scheduler support

* Quality assurance updates

* Added list support, updated testing, additional quality assurance

* Update documentation, remove extra comments and debugging

* Restructure get_role_information, format constants

* Update MCAD and Volcano list function, list tests, and prefix change

* Restore kubernetes_mcad_scheduler_test.py

* Update prefix in test cases

* Update port for Kubernetes service

* Update list, list tests, and quality assurance

* remove debugging option

* pyre fix

* pyre fixes

* Update list_failure_tests

* Update test_list_failure

* Minor formatting changes

* Additional CI tests

---------

Co-authored-by: Sara Kokkila Schumacher <Sara.Ilane.Ladd.Kokkila.Schumacher@ibm.com>

scripts/component_integration_tests.py

@@ -51,7 +51,14 @@ def main() -> None:
     torchx_image = "dummy_image"
     dryrun = False
 
-    if scheduler in ("kubernetes", "local_docker", "aws_batch", "lsf", "gcp_batch"):
+    if scheduler in (
+        "kubernetes",
+        "kubernetes_mcad",
+        "local_docker",
+        "aws_batch",
+        "lsf",
+        "gcp_batch",
+    ):
         try:
             build = build_and_push_image()
             torchx_image = build.torchx_image
@@ -71,6 +78,16 @@ def main() -> None:
                 "queue": "default",
             },
         },
+        "kubernetes_mcad": {
+            "providers": [
+                component_provider,
+                examples_app_defs_providers,
+            ],
+            "image": torchx_image,
+            "cfg": {
+                "namespace": "torchx-dev",
+            },
+        },
         "local_cwd": {
             "providers": [
                 component_provider,

torchx/schedulers/__init__.py

@@ -17,6 +17,7 @@
     "local_cwd": "torchx.schedulers.local_scheduler",
     "slurm": "torchx.schedulers.slurm_scheduler",
     "kubernetes": "torchx.schedulers.kubernetes_scheduler",
+    "kubernetes_mcad": "torchx.schedulers.kubernetes_mcad_scheduler",
     "aws_batch": "torchx.schedulers.aws_batch_scheduler",
     "gcp_batch": "torchx.schedulers.gcp_batch_scheduler",
     "ray": "torchx.schedulers.ray_scheduler",