Fix bug in violating Kubernetes naming rules

src/cloudai/_core/json_gen_strategy.py

@@ -14,6 +14,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import re
 from abc import abstractmethod
 from typing import Any, Dict
 
@@ -28,6 +29,27 @@ class JsonGenStrategy(TestTemplateStrategy):
     It specifies how to generate JSON job specifications based on system and test parameters.
     """
 
+    def sanitize_k8s_job_name(self, job_name: str) -> str:
+        """
+        Sanitize the job name to ensure it follows Kubernetes naming rules.
+
+        - Must be lowercase.
+        - Can only contain alphanumeric characters, hyphens, and periods.
+        - Must start and end with an alphanumeric character.
+        - Must be at most 253 characters long.
+
+        Args:
+            job_name (str): The original job name to be sanitized.
+
+        Returns:
+            str: The sanitized job name that complies with Kubernetes naming rules.
+        """
+        sanitized_name = job_name.lower()
+        sanitized_name = re.sub(r"[^a-z0-9.-]", "-", sanitized_name)
+        sanitized_name = re.sub(r"^[^a-z0-9]+", "", sanitized_name)
+        sanitized_name = re.sub(r"[^a-z0-9]+$", "", sanitized_name)
+        return sanitized_name[:253]
+
     @abstractmethod
     def gen_json(self, tr: TestRun) -> Dict[Any, Any]:
         """

src/cloudai/schema/test_template/nccl_test/kubernetes_json_gen_strategy.py

@@ -26,8 +26,9 @@ def gen_json(self, tr: TestRun) -> Dict[Any, Any]:
         final_env_vars = self._override_env_vars(self.system.global_env_vars, tr.test.extra_env_vars)
         final_cmd_args = self._override_cmd_args(self.default_cmd_args, tr.test.cmd_args)
         final_num_nodes = self._determine_num_nodes(tr.num_nodes, tr.nodes)
+        sanitized_job_name = self.sanitize_k8s_job_name("nccl-test")
         job_spec = self._create_job_spec(
-            tr.name, final_num_nodes, tr.nodes, final_env_vars, final_cmd_args, tr.test.extra_cmd_args
+            sanitized_job_name, final_num_nodes, tr.nodes, final_env_vars, final_cmd_args, tr.test.extra_cmd_args
         )
 
         return job_spec