skypilot-nightly 1.0.0.dev20241005__tar.gz → 1.0.0.dev20241007__tar.gz

{skypilot_nightly-1.0.0.dev20241005/skypilot_nightly.egg-info → skypilot_nightly-1.0.0.dev20241007}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: skypilot-nightly
-Version: 1.0.0.dev20241005
+Version: 1.0.0.dev20241007
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0
@@ -163,7 +163,7 @@ Requires-Dist: pyvmomi==8.0.1.0.2; extra == "all"
 
 ----
 :fire: *News* :fire:
-- [Sep, 2024] Point, Launch and Serve **Llama 3.2** on on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
+- [Sep, 2024] Point, Launch and Serve **Llama 3.2** on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
 - [Sep, 2024] Run and deploy [**Pixtral**](./llm/pixtral), the first open-source multimodal model from Mistral AI.
 - [Jul, 2024] [**Finetune**](./llm/llama-3_1-finetuning/) and [**serve**](./llm/llama-3_1/) **Llama 3.1** on your infra
 - [Jun, 2024] Reproduce **GPT** with [llm.c](https://github.com/karpathy/llm.c/discussions/481) on any cloud: [**guide**](./llm/gpt-2/)

{skypilot_nightly-1.0.0.dev20241005 → skypilot_nightly-1.0.0.dev20241007}/README.md

@@ -26,7 +26,7 @@
 
 ----
 :fire: *News* :fire:
-- [Sep, 2024] Point, Launch and Serve **Llama 3.2** on on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
+- [Sep, 2024] Point, Launch and Serve **Llama 3.2** on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
 - [Sep, 2024] Run and deploy [**Pixtral**](./llm/pixtral), the first open-source multimodal model from Mistral AI.
 - [Jul, 2024] [**Finetune**](./llm/llama-3_1-finetuning/) and [**serve**](./llm/llama-3_1/) **Llama 3.1** on your infra
 - [Jun, 2024] Reproduce **GPT** with [llm.c](https://github.com/karpathy/llm.c/discussions/481) on any cloud: [**guide**](./llm/gpt-2/)

{skypilot_nightly-1.0.0.dev20241005 → skypilot_nightly-1.0.0.dev20241007}/sky/__init__.py

@@ -5,7 +5,7 @@ from typing import Optional
 import urllib.request
 
 # Replaced with the current commit when building the wheels.
-_SKYPILOT_COMMIT_SHA = '81e19038efe6988ca4d3e78dfb24140c587fa32c'
+_SKYPILOT_COMMIT_SHA = 'd5b6d89c83ea1ee7258f68314da4c6f8add83e04'
 
 
 def _get_git_commit():
@@ -35,7 +35,7 @@ def _get_git_commit():
 
 
 __commit__ = _get_git_commit()
-__version__ = '1.0.0.dev20241005'
+__version__ = '1.0.0.dev20241007'
 __root_dir__ = os.path.dirname(os.path.abspath(__file__))
 
 

{skypilot_nightly-1.0.0.dev20241005 → skypilot_nightly-1.0.0.dev20241007}/sky/serve/controller.py

@@ -2,6 +2,7 @@
 
 Responsible for autoscaling and replica management.
 """
+import contextlib
 import logging
 import threading
 import time
@@ -9,6 +10,7 @@ import traceback
 from typing import Any, Dict, List
 
 import fastapi
+from fastapi import responses
 import uvicorn
 
 from sky import serve
@@ -49,7 +51,14 @@ class SkyServeController:
             autoscalers.Autoscaler.from_spec(service_name, service_spec))
         self._host = host
         self._port = port
-        self._app = fastapi.FastAPI()
+        self._app = fastapi.FastAPI(lifespan=self.lifespan)
+
+    @contextlib.asynccontextmanager
+    async def lifespan(self, _: fastapi.FastAPI):
+        uvicorn_access_logger = logging.getLogger('uvicorn.access')
+        for handler in uvicorn_access_logger.handlers:
+            handler.setFormatter(sky_logging.FORMATTER)
+        yield
 
     def _run_autoscaler(self):
         logger.info('Starting autoscaler.')
@@ -88,7 +97,8 @@ class SkyServeController:
     def run(self) -> None:
 
         @self._app.post('/controller/load_balancer_sync')
-        async def load_balancer_sync(request: fastapi.Request):
+        async def load_balancer_sync(
+                request: fastapi.Request) -> fastapi.Response:
             request_data = await request.json()
             # TODO(MaoZiming): Check aggregator type.
             request_aggregator: Dict[str, Any] = request_data.get(
@@ -96,18 +106,21 @@ class SkyServeController:
             timestamps: List[int] = request_aggregator.get('timestamps', [])
             logger.info(f'Received {len(timestamps)} inflight requests.')
             self._autoscaler.collect_request_information(request_aggregator)
-            return {
+            return responses.JSONResponse(content={
                 'ready_replica_urls':
                     self._replica_manager.get_active_replica_urls()
-            }
+            },
+                                          status_code=200)
 
         @self._app.post('/controller/update_service')
-        async def update_service(request: fastapi.Request):
+        async def update_service(request: fastapi.Request) -> fastapi.Response:
             request_data = await request.json()
             try:
                 version = request_data.get('version', None)
                 if version is None:
-                    return {'message': 'Error: version is not specified.'}
+                    return responses.JSONResponse(
+                        content={'message': 'Error: version is not specified.'},
+                        status_code=400)
                 update_mode_str = request_data.get(
                     'mode', serve_utils.DEFAULT_UPDATE_MODE.value)
                 update_mode = serve_utils.UpdateMode(update_mode_str)
@@ -136,17 +149,13 @@ class SkyServeController:
                 self._autoscaler.update_version(version,
                                                 service,
                                                 update_mode=update_mode)
-                return {'message': 'Success'}
+                return responses.JSONResponse(content={'message': 'Success'},
+                                              status_code=200)
             except Exception as e:  # pylint: disable=broad-except
                 logger.error(f'Error in update_service: '
                              f'{common_utils.format_exception(e)}')
-                return {'message': 'Error'}
-
-        @self._app.on_event('startup')
-        def configure_logger():
-            uvicorn_access_logger = logging.getLogger('uvicorn.access')
-            for handler in uvicorn_access_logger.handlers:
-                handler.setFormatter(sky_logging.FORMATTER)
+                return responses.JSONResponse(content={'message': 'Error'},
+                                              status_code=500)
 
         threading.Thread(target=self._run_autoscaler).start()
 

{skypilot_nightly-1.0.0.dev20241005 → skypilot_nightly-1.0.0.dev20241007}/sky/serve/serve_utils.py

@@ -302,6 +302,10 @@ def update_service_encoded(service_name: str, version: int, mode: str) -> str:
         raise ValueError('The service is up-ed in an old version and does not '
                          'support update. Please `sky serve down` '
                          'it first and relaunch the service. ')
+    elif resp.status_code == 400:
+        raise ValueError(f'Client error during service update: {resp.text}')
+    elif resp.status_code == 500:
+        raise RuntimeError(f'Server error during service update: {resp.text}')
     elif resp.status_code != 200:
         raise ValueError(f'Failed to update service: {resp.text}')
 

{skypilot_nightly-1.0.0.dev20241005 → skypilot_nightly-1.0.0.dev20241007}/sky/utils/subprocess_utils.py

@@ -77,6 +77,7 @@ def handle_returncode(returncode: int,
         command: The command that was run.
         error_msg: The error message to print.
         stderr: The stderr of the command.
+        stream_logs: Whether to stream logs.
     """
     echo = logger.error if stream_logs else logger.debug
     if returncode != 0:

{skypilot_nightly-1.0.0.dev20241005 → skypilot_nightly-1.0.0.dev20241007/skypilot_nightly.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: skypilot-nightly
-Version: 1.0.0.dev20241005
+Version: 1.0.0.dev20241007
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0
@@ -163,7 +163,7 @@ Requires-Dist: pyvmomi==8.0.1.0.2; extra == "all"
 
 ----
 :fire: *News* :fire:
-- [Sep, 2024] Point, Launch and Serve **Llama 3.2** on on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
+- [Sep, 2024] Point, Launch and Serve **Llama 3.2** on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
 - [Sep, 2024] Run and deploy [**Pixtral**](./llm/pixtral), the first open-source multimodal model from Mistral AI.
 - [Jul, 2024] [**Finetune**](./llm/llama-3_1-finetuning/) and [**serve**](./llm/llama-3_1/) **Llama 3.1** on your infra
 - [Jun, 2024] Reproduce **GPT** with [llm.c](https://github.com/karpathy/llm.c/discussions/481) on any cloud: [**guide**](./llm/gpt-2/)