skypilot-nightly 1.0.0.dev20250327__tar.gz → 1.0.0.dev20250329__tar.gz

{skypilot_nightly-1.0.0.dev20250327/skypilot_nightly.egg-info → skypilot_nightly-1.0.0.dev20250329}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: skypilot-nightly
-Version: 1.0.0.dev20250327
+Version: 1.0.0.dev20250329
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0
@@ -230,7 +230,7 @@ SkyPilot **is easy to use for AI users**:
 - Easy job management: queue, run, and auto-recover many jobs
 
 SkyPilot **unifies multiple clusters, clouds, and hardware**:
-- One interface to use reserved GPUs, Kubernetes clusters, or 15+ clouds
+- One interface to use reserved GPUs, Kubernetes clusters, or 16+ clouds
 - [Flexible provisioning](https://docs.skypilot.co/en/latest/examples/auto-failover.html) of GPUs, TPUs, CPUs, with auto-retry
 - [Team deployment](https://docs.skypilot.co/en/latest/reference/api-server/api-server.html) and resource sharing
 
@@ -262,6 +262,7 @@ VMware vSphere, Nebius.
 <p align="center">
   <img alt="SkyPilot" src="https://raw.githubusercontent.com/skypilot-org/skypilot/master/docs/source/images/cloud-logos-light.png" width=85%>
 </p>
+<!-- source xcf file: https://drive.google.com/drive/folders/1S_acjRsAD3T14qMeEnf6FFrIwHu_Gs_f?usp=drive_link -->
 
 
 ## Getting started

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/README.md

@@ -59,7 +59,7 @@ SkyPilot **is easy to use for AI users**:
 - Easy job management: queue, run, and auto-recover many jobs
 
 SkyPilot **unifies multiple clusters, clouds, and hardware**:
-- One interface to use reserved GPUs, Kubernetes clusters, or 15+ clouds
+- One interface to use reserved GPUs, Kubernetes clusters, or 16+ clouds
 - [Flexible provisioning](https://docs.skypilot.co/en/latest/examples/auto-failover.html) of GPUs, TPUs, CPUs, with auto-retry
 - [Team deployment](https://docs.skypilot.co/en/latest/reference/api-server/api-server.html) and resource sharing
 
@@ -94,6 +94,7 @@ VMware vSphere, Nebius.
     <img alt="SkyPilot" src="https://raw.githubusercontent.com/skypilot-org/skypilot/master/docs/source/images/cloud-logos-light.png" width=85%>
   </picture>
 </p>
+<!-- source xcf file: https://drive.google.com/drive/folders/1S_acjRsAD3T14qMeEnf6FFrIwHu_Gs_f?usp=drive_link -->
 
 
 ## Getting started

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/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 = '3618e2b85057a16b6b8204fd69d450faf4eb9dd2'
+_SKYPILOT_COMMIT_SHA = '9e6cef22ae679a1fef4539cbfc02a6d5cdc405ed'
 
 
 def _get_git_commit():
@@ -35,7 +35,7 @@ def _get_git_commit():
 
 
 __commit__ = _get_git_commit()
-__version__ = '1.0.0.dev20250327'
+__version__ = '1.0.0.dev20250329'
 __root_dir__ = os.path.dirname(os.path.abspath(__file__))
 
 

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/sky/clouds/do.py

@@ -280,6 +280,8 @@ class DO(clouds.Cloud):
         return True, None
 
     def get_credential_file_mounts(self) -> Dict[str, str]:
+        if do_utils.CREDENTIALS_PATH is None:
+            return {}
         if not os.path.exists(os.path.expanduser(do_utils.CREDENTIALS_PATH)):
             return {}
         return {

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/sky/optimizer.py

@@ -713,7 +713,7 @@ class Optimizer:
         if message_data:
             metric = 'COST ($)' if minimize_cost else 'TIME (s)'
             table = _create_table(['SOURCE', 'TARGET', 'SIZE (GB)', metric])
-            table.add_rows(reversed(message_data))
+            table.add_rows(list(reversed(message_data)))
             logger.info(f'Egress plan:\n{table}\n')
 
     @staticmethod

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/sky/server/common.py

@@ -51,6 +51,11 @@ API_SERVER_CMD = '-m sky.server.server'
 API_SERVER_CLIENT_DIR = pathlib.Path('~/.sky/api_server/clients')
 RETRY_COUNT_ON_TIMEOUT = 3
 
+# The maximum time to wait for the API server to start, set to a conservative
+# value that unlikely to reach since the server might be just starting slowly
+# (e.g. in high contention env) and we will exit eagerly if server exit.
+WAIT_APISERVER_START_TIMEOUT_SEC = 60
+
 SKY_API_VERSION_WARNING = (
     f'{colorama.Fore.YELLOW}SkyPilot API server is too old: '
     f'v{{server_version}} (client version is v{{client_version}}). '
@@ -179,7 +184,8 @@ def _start_api_server(deploy: bool = False,
     server_url = get_server_url(host)
     assert server_url in AVAILABLE_LOCAL_API_SERVER_URLS, (
         f'server url {server_url} is not a local url')
-    with rich_utils.client_status('Starting SkyPilot API server'):
+    with rich_utils.client_status('Starting SkyPilot API server, '
+                                  f'view logs at {constants.API_SERVER_LOGS}'):
         logger.info(f'{colorama.Style.DIM}Failed to connect to '
                     f'SkyPilot API server at {server_url}. '
                     'Starting a local server.'
@@ -216,14 +222,16 @@ def _start_api_server(deploy: bool = False,
         # If this is called from a CLI invocation, we need
         # start_new_session=True so that SIGINT on the CLI will not also kill
         # the API server.
-        subprocess.Popen(cmd, shell=True, start_new_session=True)
+        proc = subprocess.Popen(cmd, shell=True, start_new_session=True)
 
-        # Wait for the server to start until timeout.
-        # Conservative upper time bound for starting the server based on
-        # profiling.
-        timeout_sec = 12
         start_time = time.time()
         while True:
+            # Check if process has exited
+            if proc.poll() is not None:
+                with ux_utils.print_exception_no_traceback():
+                    raise RuntimeError(
+                        'SkyPilot API server process exited unexpectedly.\n'
+                        f'View logs at: {constants.API_SERVER_LOGS}')
             api_server_info = get_api_server_status()
             assert api_server_info.status != ApiServerStatus.VERSION_MISMATCH, (
                 f'API server version mismatch when starting the server. '
@@ -231,7 +239,7 @@ def _start_api_server(deploy: bool = False,
                 f'Client version: {server_constants.API_VERSION}')
             if api_server_info.status == ApiServerStatus.HEALTHY:
                 break
-            elif time.time() - start_time >= timeout_sec:
+            elif time.time() - start_time >= WAIT_APISERVER_START_TIMEOUT_SEC:
                 with ux_utils.print_exception_no_traceback():
                     raise RuntimeError(
                         'Failed to start SkyPilot API server at '

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/sky/server/requests/executor.py

@@ -465,7 +465,7 @@ def start(deploy: bool) -> List[multiprocessing.Process]:
             target=mp_queue.start_queue_manager, args=(queue_names, port))
         queue_server.start()
         sub_procs.append(queue_server)
-        mp_queue.wait_for_queues_to_be_ready(queue_names, port=port)
+        mp_queue.wait_for_queues_to_be_ready(queue_names, queue_server, port)
 
     logger.info('Request queues created')
 

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/sky/server/requests/queues/mp_queue.py

@@ -1,4 +1,5 @@
 """Shared queues for multiprocessing."""
+import multiprocessing
 from multiprocessing import managers
 import queue
 import time
@@ -57,10 +58,13 @@ def get_queue(queue_name: str,
 
 
 def wait_for_queues_to_be_ready(queue_names: List[str],
+                                queue_server: multiprocessing.Process,
                                 port: int = DEFAULT_QUEUE_MANAGER_PORT) -> None:
     """Wait for the queues to be ready after queue manager is just started."""
     initial_time = time.time()
-    max_wait_time = 5
+    # Wait for queue manager to be ready. Exit eagerly if the manager process
+    # exits, just wait for a long timeout that is unlikely to reach otherwise.
+    max_wait_time = 60
     while queue_names:
         try:
             get_queue(queue_names[0], port)
@@ -70,6 +74,9 @@ def wait_for_queues_to_be_ready(queue_names: List[str],
             logger.info(f'Waiting for request queue, named {queue_names[0]!r}, '
                         f'to be ready...')
             time.sleep(0.2)
+            if not queue_server.is_alive():
+                raise RuntimeError(
+                    'Queue manager process exited unexpectedly.') from e
             if time.time() - initial_time > max_wait_time:
                 raise RuntimeError(
                     f'Request queue, named {queue_names[0]!r}, '

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329}/sky/server/server.py

@@ -1116,7 +1116,7 @@ if __name__ == '__main__':
 
     sub_procs = []
     try:
-        sub_procs = executor.start(cmd_args.deploy)
+        sub_procs = executor.start(deploy=cmd_args.deploy)
         logger.info(f'Starting SkyPilot API server, workers={num_workers}')
         # We don't support reload for now, since it may cause leakage of request
         # workers or interrupt running requests.

{skypilot_nightly-1.0.0.dev20250327 → skypilot_nightly-1.0.0.dev20250329/skypilot_nightly.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: skypilot-nightly
-Version: 1.0.0.dev20250327
+Version: 1.0.0.dev20250329
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0
@@ -230,7 +230,7 @@ SkyPilot **is easy to use for AI users**:
 - Easy job management: queue, run, and auto-recover many jobs
 
 SkyPilot **unifies multiple clusters, clouds, and hardware**:
-- One interface to use reserved GPUs, Kubernetes clusters, or 15+ clouds
+- One interface to use reserved GPUs, Kubernetes clusters, or 16+ clouds
 - [Flexible provisioning](https://docs.skypilot.co/en/latest/examples/auto-failover.html) of GPUs, TPUs, CPUs, with auto-retry
 - [Team deployment](https://docs.skypilot.co/en/latest/reference/api-server/api-server.html) and resource sharing
 
@@ -262,6 +262,7 @@ VMware vSphere, Nebius.
 <p align="center">
   <img alt="SkyPilot" src="https://raw.githubusercontent.com/skypilot-org/skypilot/master/docs/source/images/cloud-logos-light.png" width=85%>
 </p>
+<!-- source xcf file: https://drive.google.com/drive/folders/1S_acjRsAD3T14qMeEnf6FFrIwHu_Gs_f?usp=drive_link -->
 
 
 ## Getting started