[Core] Revert "Update --labels and add --labels-file options for Label Selector API (#51706)

java/test/src/main/java/io/ray/test/NodeLabelSchedulingTest.java

@@ -24,7 +24,7 @@ public void testEmptyNodeLabels() {
   }
 
   public void testSetNodeLabels() {
-    System.setProperty("ray.head-args.0", "--labels=\"gpu_type=A100,azone=azone-1\"");
+    System.setProperty("ray.head-args.0", "--labels={\"gpu_type\":\"A100\",\"azone\":\"azone-1\"}");
     try {
       Ray.init();
       List<NodeInfo> nodeInfos = Ray.getRuntimeContext().getAllNodeInfo();

python/ray/_private/parameter.py

@@ -3,8 +3,10 @@
 from typing import Dict, List, Optional
 
 import ray._private.ray_constants as ray_constants
-from ray._private.label_utils import validate_node_labels
-from ray._private.utils import check_ray_client_dependencies_installed
+from ray._private.utils import (
+    validate_node_labels,
+    check_ray_client_dependencies_installed,
+)
 
 
 logger = logging.getLogger(__name__)

python/ray/_private/utils.py

@@ -1803,6 +1803,43 @@ def update_envs(env_vars: Dict[str, str]):
         os.environ[key] = result
 
 
+def parse_node_labels_json(
+    labels_json: str, cli_logger, cf, command_arg="--labels"
+) -> Dict[str, str]:
+    try:
+        labels = json.loads(labels_json)
+        if not isinstance(labels, dict):
+            raise ValueError(
+                "The format after deserialization is not a key-value pair map"
+            )
+        for key, value in labels.items():
+            if not isinstance(key, str):
+                raise ValueError("The key is not string type.")
+            if not isinstance(value, str):
+                raise ValueError(f'The value of the "{key}" is not string type')
+    except Exception as e:
+        cli_logger.abort(
+            "`{}` is not a valid JSON string, detail error:{}"
+            "Valid values look like this: `{}`",
+            cf.bold(f"{command_arg}={labels_json}"),
+            str(e),
+            cf.bold(f'{command_arg}=\'{{"gpu_type": "A100", "region": "us"}}\''),
+        )
+    return labels
+
+
+def validate_node_labels(labels: Dict[str, str]):
+    if labels is None:
+        return
+    for key in labels.keys():
+        if key.startswith(ray_constants.RAY_DEFAULT_LABEL_KEYS_PREFIX):
+            raise ValueError(
+                f"Custom label keys `{key}` cannot start with the prefix "
+                f"`{ray_constants.RAY_DEFAULT_LABEL_KEYS_PREFIX}`. "
+                f"This is reserved for Ray defined labels."
+            )
+
+
 def parse_pg_formatted_resources_to_original(
     pg_formatted_resources: Dict[str, float]
 ) -> Dict[str, float]:

python/ray/scripts/scripts.py

@@ -25,14 +25,11 @@
 import ray
 import ray._private.ray_constants as ray_constants
 import ray._private.services as services
-from ray._private.label_utils import (
-    parse_node_labels_from_yaml_file,
-    parse_node_labels_string,
-)
 from ray._private.utils import (
     check_ray_client_dependencies_installed,
     load_class,
     parse_resources_json,
+    parse_node_labels_json,
 )
 from ray._private.internal_api import memory_summary
 from ray._private.usage import usage_lib
@@ -638,19 +635,9 @@ def debug(address: str, verbose: bool):
     "--labels",
     required=False,
     hidden=True,
-    default="",
-    type=str,
-    help="a string list of key-value pairs mapping label name to label value."
-    "These values take precedence over conflicting keys passed in from --labels-file."
-    'Ex: --labels "key1=val1,key2=val2"',
-)
-@click.option(
-    "--labels-file",
-    required=False,
-    hidden=True,
-    default="",
+    default="{}",
     type=str,
-    help="a path to a YAML file containing a dictionary mapping of label keys to values.",
+    help="a JSON serialized dictionary mapping label name to label value.",
 )
 @click.option(
     "--include-log-monitor",
@@ -708,7 +695,6 @@ def start(
     ray_debugger_external,
     disable_usage_stats,
     labels,
-    labels_file,
     include_log_monitor,
 ):
     """Start Ray processes manually on the local machine."""
@@ -729,34 +715,7 @@ def start(
         node_ip_address = services.resolve_ip_for_localhost(node_ip_address)
 
     resources = parse_resources_json(resources, cli_logger, cf)
-
-    # Compose labels passed in with `--labels` and `--labels-file`.
-    # The label value from `--labels` will overrwite the value of any duplicate keys.
-    try:
-        labels_from_file_dict = parse_node_labels_from_yaml_file(labels_file)
-    except Exception as e:
-        cli_logger.abort(
-            "The file at `{}` is not a valid YAML file, detailed error:{}"
-            "Valid values look like this: `{}`",
-            cf.bold(f"--labels-file={labels_file}"),
-            str(e),
-            cf.bold("--labels-file='gpu_type: A100\nregion: us'"),
-        )
-    try:
-        labels_from_string = parse_node_labels_string(labels)
-    except Exception as e:
-        cli_logger.abort(
-            "`{}` is not a valid string of key-value pairs, detail error:{}"
-            "Valid values look like this: `{}`",
-            cf.bold(f"--labels={labels}"),
-            str(e),
-            cf.bold('--labels="key1=val1,key2=val2"'),
-        )
-    labels_dict = (
-        {**labels_from_file_dict, **labels_from_string}
-        if labels_from_file_dict
-        else labels_from_string
-    )
+    labels_dict = parse_node_labels_json(labels, cli_logger, cf)
 
     if plasma_store_socket_name is not None:
         warnings.warn(

python/ray/tests/BUILD

@@ -349,7 +349,6 @@ py_test_module_list(
 py_test_module_list(
     size = "medium",
     files = [
-        "test_label_utils.py",
         "test_minimal_install.py",
         "test_runtime_env_ray_minimal.py",
         "test_utils.py",

python/ray/tests/test_node_label_scheduling_strategy.py

@@ -31,7 +31,7 @@ def get_node_id():
 
 @pytest.mark.parametrize(
     "call_ray_start",
-    ["ray start --head --labels gpu_type=A100,region=us"],
+    ['ray start --head --labels={"gpu_type":"A100","region":"us"}'],
     indirect=True,
 )
 def test_node_label_scheduling_basic(call_ray_start):