Add schema pattern check to pytorch-job template (#178)

* Update readme for volume flag

* Add schema pattern check to pytorch-job template, unit test added, all test passed locally

Author: mollyheamazon

hyperpod-pytorch-job-template/hyperpod_pytorch_job_template/v1_0/model.py

@@ -15,11 +15,27 @@
 
 
 class VolumeConfig(BaseModel):
-    name: str = Field(..., description="Volume name")
+    name: str = Field(
+        ..., 
+        description="Volume name",
+        min_length=1
+    )
     type: Literal['hostPath', 'pvc'] = Field(..., description="Volume type")
-    mount_path: str = Field(..., description="Mount path in container")
-    path: Optional[str] = Field(None, description="Host path (required for hostPath volumes)")
-    claim_name: Optional[str] = Field(None, description="PVC claim name (required for pvc volumes)")
+    mount_path: str = Field(
+        ..., 
+        description="Mount path in container",
+        min_length=1
+    )
+    path: Optional[str] = Field(
+        None, 
+        description="Host path (required for hostPath volumes)",
+        min_length=1
+    )
+    claim_name: Optional[str] = Field(
+        None, 
+        description="PVC claim name (required for pvc volumes)",
+        min_length=1
+    )
     read_only: Optional[Literal['true', 'false']] = Field(None, description="Read-only flag for pvc volumes")
 
     @field_validator('mount_path', 'path')
@@ -47,9 +63,22 @@ def validate_type_specific_fields(self):
 class PyTorchJobConfig(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
-    job_name: str = Field(alias="job_name", description="Job name")
-    image: str = Field(description="Docker image for training")
-    namespace: Optional[str] = Field(default=None, description="Kubernetes namespace")
+    job_name: str = Field(
+        alias="job_name", 
+        description="Job name",
+        min_length=1,
+        max_length=63,
+        pattern=r'^[a-z0-9]([-a-z0-9]*[a-z0-9])?$'
+    )
+    image: str = Field(
+        description="Docker image for training",
+        min_length=1
+    )
+    namespace: Optional[str] = Field(
+        default=None, 
+        description="Kubernetes namespace",
+        min_length=1
+    )
     command: Optional[List[str]] = Field(
         default=None, description="Command to run in the container"
     )
@@ -60,16 +89,28 @@ class PyTorchJobConfig(BaseModel):
         default=None, description="Environment variables as key_value pairs"
     )
     pull_policy: Optional[str] = Field(
-        default=None, alias="pull_policy", description="Image pull policy"
+        default=None, 
+        alias="pull_policy", 
+        description="Image pull policy",
+        min_length=1
     )
     instance_type: Optional[str] = Field(
-        default=None, alias="instance_type", description="Instance type for training"
+        default=None, 
+        alias="instance_type", 
+        description="Instance type for training",
+        min_length=1
     )
     node_count: Optional[int] = Field(
-        default=None, alias="node_count", description="Number of nodes"
+        default=None, 
+        alias="node_count", 
+        description="Number of nodes",
+        ge=1
     )
     tasks_per_node: Optional[int] = Field(
-        default=None, alias="tasks_per_node", description="Number of tasks per node"
+        default=None, 
+        alias="tasks_per_node", 
+        description="Number of tasks per node",
+        ge=1
     )
     label_selector: Optional[Dict[str, str]] = Field(
         default=None,
@@ -82,16 +123,29 @@ class PyTorchJobConfig(BaseModel):
         description="Schedule pods only on nodes that passed deep health check",
     )
     scheduler_type: Optional[str] = Field(
-        default=None, alias="scheduler_type", description="Scheduler type"
+        default=None, 
+        alias="scheduler_type", 
+        description="Scheduler type",
+        min_length=1
     )
     queue_name: Optional[str] = Field(
-        default=None, alias="queue_name", description="Queue name for job scheduling"
+        default=None, 
+        alias="queue_name", 
+        description="Queue name for job scheduling",
+        min_length=1,
+        max_length=63,
+        pattern=r'^[a-z0-9]([-a-z0-9]*[a-z0-9])?$'
     )
     priority: Optional[str] = Field(
-        default=None, description="Priority class for job scheduling"
+        default=None, 
+        description="Priority class for job scheduling",
+        min_length=1
     )
     max_retry: Optional[int] = Field(
-        default=None, alias="max_retry", description="Maximum number of job retries"
+        default=None, 
+        alias="max_retry", 
+        description="Maximum number of job retries",
+        ge=0
     )
     volume: Optional[List[VolumeConfig]] = Field(
         default=None, description="List of volume configurations. \
@@ -102,7 +156,10 @@ class PyTorchJobConfig(BaseModel):
         "
     )
     service_account_name: Optional[str] = Field(
-        default=None, alias="service_account_name", description="Service account name"
+        default=None, 
+        alias="service_account_name", 
+        description="Service account name",
+        min_length=1
     )
 
     @field_validator('volume')
@@ -123,6 +180,52 @@ def validate_no_duplicates(cls, v):
 
         return v
 
+    @field_validator('command', 'args')
+    def validate_string_lists(cls, v):
+        """Validate that command and args contain non-empty strings."""
+        if not v:
+            return v
+        
+        for i, item in enumerate(v):
+            if not isinstance(item, str) or not item.strip():
+                field_name = cls.model_fields.get('command', {}).get('alias', 'command') if 'command' in str(v) else 'args'
+                raise ValueError(f"{field_name}[{i}] must be a non-empty string")
+        
+        return v
+
+    @field_validator('environment')
+    def validate_environment_variable_names(cls, v):
+        """Validate environment variable names follow C_IDENTIFIER pattern."""
+        if not v:
+            return v
+        
+        import re
+        c_identifier_pattern = re.compile(r'^[a-zA-Z_][a-zA-Z0-9_]*$')
+        
+        for key in v.keys():
+            if not c_identifier_pattern.match(key):
+                raise ValueError(f"Environment variable name '{key}' must be a valid C_IDENTIFIER")
+        
+        return v
+
+    @field_validator('label_selector')
+    def validate_label_selector_keys(cls, v):
+        """Validate label selector keys follow Kubernetes label naming conventions."""
+        if not v:
+            return v
+        
+        import re
+        # Kubernetes label key pattern - allows namespaced labels like kubernetes.io/arch
+        # Pattern: [prefix/]name where prefix and name follow DNS subdomain rules
+        # Also reject double dots
+        label_key_pattern = re.compile(r'^([a-zA-Z0-9]([a-zA-Z0-9\-_.]*[a-zA-Z0-9])?/)?[a-zA-Z0-9]([a-zA-Z0-9\-_.]*[a-zA-Z0-9])?$')
+        
+        for key in v.keys():
+            if not key or not label_key_pattern.match(key) or '..' in key:
+                raise ValueError(f"Label selector key '{key}' must follow Kubernetes label naming conventions")
+        
+        return v
+
     def to_domain(self) -> Dict:
         """
         Convert flat config to domain model (HyperPodPytorchJobSpec)

hyperpod-pytorch-job-template/hyperpod_pytorch_job_template/v1_0/schema.json

@@ -4,6 +4,7 @@
       "properties": {
         "name": {
           "description": "Volume name",
+          "minLength": 1,
           "title": "Name",
           "type": "string"
         },
@@ -18,12 +19,14 @@
         },
         "mount_path": {
           "description": "Mount path in container",
+          "minLength": 1,
           "title": "Mount Path",
           "type": "string"
         },
         "path": {
           "anyOf": [
             {
+              "minLength": 1,
               "type": "string"
             },
             {
@@ -37,6 +40,7 @@
         "claim_name": {
           "anyOf": [
             {
+              "minLength": 1,
               "type": "string"
             },
             {
@@ -78,17 +82,22 @@
   "properties": {
     "job_name": {
       "description": "Job name",
+      "maxLength": 63,
+      "minLength": 1,
+      "pattern": "^[a-z0-9]([-a-z0-9]*[a-z0-9])?$",
       "title": "Job Name",
       "type": "string"
     },
     "image": {
       "description": "Docker image for training",
+      "minLength": 1,
       "title": "Image",
       "type": "string"
     },
     "namespace": {
       "anyOf": [
         {
+          "minLength": 1,
           "type": "string"
         },
         {
@@ -150,6 +159,7 @@
     "pull_policy": {
       "anyOf": [
         {
+          "minLength": 1,
           "type": "string"
         },
         {
@@ -163,6 +173,7 @@
     "instance_type": {
       "anyOf": [
         {
+          "minLength": 1,
           "type": "string"
         },
         {
@@ -176,6 +187,7 @@
     "node_count": {
       "anyOf": [
         {
+          "minimum": 1,
           "type": "integer"
         },
         {
@@ -189,6 +201,7 @@
     "tasks_per_node": {
       "anyOf": [
         {
+          "minimum": 1,
           "type": "integer"
         },
         {
@@ -231,6 +244,7 @@
     "scheduler_type": {
       "anyOf": [
         {
+          "minLength": 1,
           "type": "string"
         },
         {
@@ -244,6 +258,9 @@
     "queue_name": {
       "anyOf": [
         {
+          "maxLength": 63,
+          "minLength": 1,
+          "pattern": "^[a-z0-9]([-a-z0-9]*[a-z0-9])?$",
           "type": "string"
         },
         {
@@ -257,6 +274,7 @@
     "priority": {
       "anyOf": [
         {
+          "minLength": 1,
           "type": "string"
         },
         {
@@ -270,6 +288,7 @@
     "max_retry": {
       "anyOf": [
         {
+          "minimum": 0,
           "type": "integer"
         },
         {
@@ -299,6 +318,7 @@
     "service_account_name": {
       "anyOf": [
         {
+          "minLength": 1,
           "type": "string"
         },
         {

src/sagemaker/hyperpod/common/config/metadata.py

@@ -6,13 +6,13 @@ class Metadata(BaseModel):
     """Metadata class"""
 
     name: str = Field(
-        description="Name must match the name of one entry in pod.spec.resourceClaims of the Pod where this field is used. It makes that resource available inside a container."
+        description="The name of the Kubernetes resource. Must follow RFC1123 naming conventions: lowercase alphanumeric characters or hyphens, start and end with alphanumeric character, 1-63 characters long (e.g., 'my-pytorch-job-123')."
     )
     namespace: Optional[str] = Field(
         default=None,
-        description="Name must match the name of one entry in pod.spec.resourceClaims of the Pod where this field is used. It makes that resource available inside a container.",
+        description="The Kubernetes namespace where the resource will be created. If not specified, uses the default namespace or the namespace configured in your cluster context.",
     )
     labels: Optional[Dict[str, str]] = Field(
         default=None,
-        description="Labels are key value pairs that are attached to objects, such as Pod. Labels are intended to be used to specify identifying attributes of objects. The system ignores labels that are not in the service's selector. Labels can only be added to objects during creation. More info: XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX",
+        description="Labels are key value pairs that are attached to objects, such as Pod. Labels are intended to be used to specify identifying attributes of objects. The system ignores labels that are not in the service's selector. Labels can only be added to objects during creation.",
     )