fix: exceed payload size limit as invocation error

Throw CheckpointError with error_category as INVOCATION for 4xx
InvalidParameterValueException errors related to payload size limit
exceeded, as these errors are not retryable.

Author: andreas-baakind-adsk

src/aws_durable_execution_sdk_python/exceptions.py

@@ -170,9 +170,14 @@ def from_exception(cls, exception: Exception) -> CheckpointError:
             and (
                 # is not InvalidParam => Execution
                 (error.get("Code", "") or "") != "InvalidParameterValueException"
-                # is not Invalid Token => Execution
-                or not (error.get("Message") or "").startswith(
-                    "Invalid Checkpoint Token"
+                or not (
+                    # is not Invalid Token => Execution
+                    (error.get("Message") or "").startswith("Invalid Checkpoint Token")
+                    or
+                    # is not Output Payload Too Large => Execution
+                    (error.get("Message") or "").startswith(
+                        "STEP output payload size must be less than or equal to"
+                    )
                 )
             )
         ):

tests/exceptions_test.py

@@ -70,6 +70,23 @@ def test_checkpoint_error_classification_invalid_token_invocation():
     assert not result.is_retriable()
 
 
+def test_checkpoint_error_classification_payload_size_exceeded_invocation():
+    """Test 4xx InvalidParameterValueException with STEP output payload size limit exceeded is invocation error."""
+    error_response = {
+        "Error": {
+            "Code": "InvalidParameterValueException",
+            "Message": "STEP output payload size must be less than or equal to 262144 bytes.",
+        },
+        "ResponseMetadata": {"HTTPStatusCode": 400},
+    }
+    client_error = ClientError(error_response, "Checkpoint")
+
+    result = CheckpointError.from_exception(client_error)
+
+    assert result.error_category == CheckpointErrorCategory.INVOCATION
+    assert not result.is_retriable()
+
+
 def test_checkpoint_error_classification_other_4xx_execution():
     """Test other 4xx errors are execution errors."""
     error_response = {