Merge branch 'main' into feat/k8s_pod_overlay

Author: clumsy

scripts/setup_pyre.sh

@@ -822,16 +822,23 @@ def _run_opts(self) -> runopts:
         return opts
 
     def describe(self, app_id: str) -> Optional[DescribeAppResponse]:
+        from kubernetes.client.rest import ApiException
+
         namespace, name = app_id.split(":")
         roles = {}
         roles_statuses = {}
-        resp = self._custom_objects_api().get_namespaced_custom_object_status(
-            group="batch.volcano.sh",
-            version="v1alpha1",
-            namespace=namespace,
-            plural="jobs",
-            name=name,
-        )
+        try:
+            resp = self._custom_objects_api().get_namespaced_custom_object_status(
+                group="batch.volcano.sh",
+                version="v1alpha1",
+                namespace=namespace,
+                plural="jobs",
+                name=name,
+            )
+        except ApiException as e:
+            if e.status == 404:
+                return None
+            raise
         status = resp.get("status")
         if status:
             state_str = status["state"]["phase"]

torchx/schedulers/kubernetes_scheduler.py

@@ -714,6 +714,32 @@ def test_describe_unknown(
             ),
         )
 
+    @patch("kubernetes.client.CustomObjectsApi.get_namespaced_custom_object_status")
+    def test_describe_api_exception_404(
+        self, get_namespaced_custom_object_status: MagicMock
+    ) -> None:
+        from kubernetes.client.rest import ApiException
+
+        api_exc = ApiException(status=404, reason="Not Found")
+        get_namespaced_custom_object_status.side_effect = api_exc
+        app_id = "testnamespace:testid"
+        scheduler = create_scheduler("test")
+        info = scheduler.describe(app_id)
+        self.assertIsNone(info)
+
+    @patch("kubernetes.client.CustomObjectsApi.get_namespaced_custom_object_status")
+    def test_describe_api_exception_other(
+        self, get_namespaced_custom_object_status: MagicMock
+    ) -> None:
+        from kubernetes.client.rest import ApiException
+
+        api_exc = ApiException(status=500, reason="Internal Server Error")
+        get_namespaced_custom_object_status.side_effect = api_exc
+        app_id = "testnamespace:testid"
+        scheduler = create_scheduler("test")
+        with self.assertRaises(ApiException):
+            scheduler.describe(app_id)
+
     def test_runopts(self) -> None:
         scheduler = kubernetes_scheduler.create_scheduler("foo")
         runopts = scheduler.run_opts()