fix: Make it running (part 1)

Author: achimnol

src/ai/backend/manager/api/etcd.py

@@ -90,7 +90,8 @@ async def get_resource_slots(request: web.Request) -> web.Response:
     log.info("ETCD.GET_RESOURCE_SLOTS ()")
     root_ctx: RootContext = request.app["_root.context"]
     known_slots = await root_ctx.config_provider.legacy_etcd_config_loader.get_resource_slots()
-    return web.json_response(known_slots, status=HTTPStatus.OK)
+    serializable_known_slots = {str(k): v.value for k, v in known_slots.items()}
+    return web.json_response(serializable_known_slots, status=HTTPStatus.OK)
 
 
 @check_api_params(

src/ai/backend/manager/api/service.py

@@ -3,7 +3,6 @@
 from collections.abc import Iterable, Sequence
 from dataclasses import dataclass
 from datetime import datetime
-from decimal import Decimal
 from http import HTTPStatus
 from typing import TYPE_CHECKING, Any, Optional, Self
 
@@ -30,7 +29,6 @@
 from ai.backend.common.types import (
     MODEL_SERVICE_RUNTIME_PROFILES,
     AccessKey,
-    BinarySize,
     ClusterMode,
     ResourceSlot,
     RuntimeVariant,
@@ -375,11 +373,11 @@ class ServiceConfigModel(LegacyBaseRequestModel):
     )
     resources: ResourceSlot = Field(
         examples=[
-            ResourceSlot({
-                "cpu": Decimal(4),
-                "mem": Decimal(BinarySize.from_str("32g")),
-                "cuda.shares": Decimal("2.5"),
-            })
+            {
+                "cpu": 4,
+                "mem": "32g",
+                "cuda.shares": "2.5",
+            }
         ]
     )
     resource_opts: dict[str, str | int | bool] = Field(examples=[{"shmem": "2g"}], default={})