PyPI - skypilot-nightly - Versions diffs - 1.0.0.dev20251016__py3-none-any.whl → 1.0.0.dev20251018__py3-none-any.whl - Mend

skypilot-nightly 1.0.0.dev20251016py3-none-any.whl → 1.0.0.dev20251018py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of skypilot-nightly might be problematic. Click here for more details.

Files changed (40) hide show

sky/__init__.py +2 -2
sky/authentication.py +17 -157
sky/backends/backend_utils.py +6 -5
sky/catalog/kubernetes_catalog.py +5 -3
sky/client/cli/command.py +0 -1
sky/dashboard/out/404.html +1 -1
sky/dashboard/out/clusters/[cluster]/[job].html +1 -1
sky/dashboard/out/clusters/[cluster].html +1 -1
sky/dashboard/out/clusters.html +1 -1
sky/dashboard/out/config.html +1 -1
sky/dashboard/out/index.html +1 -1
sky/dashboard/out/infra/[context].html +1 -1
sky/dashboard/out/infra.html +1 -1
sky/dashboard/out/jobs/[job].html +1 -1
sky/dashboard/out/jobs/pools/[pool].html +1 -1
sky/dashboard/out/jobs.html +1 -1
sky/dashboard/out/users.html +1 -1
sky/dashboard/out/volumes.html +1 -1
sky/dashboard/out/workspace/new.html +1 -1
sky/dashboard/out/workspaces/[name].html +1 -1
sky/dashboard/out/workspaces.html +1 -1
sky/provision/fluidstack/instance.py +2 -2
sky/provision/seeweb/instance.py +3 -3
sky/serve/service.py +2 -2
sky/server/requests/preconditions.py +2 -2
sky/server/requests/requests.py +32 -24
sky/server/server.py +4 -5
sky/server/stream_utils.py +10 -3
sky/setup_files/dependencies.py +19 -8
sky/utils/auth_utils.py +153 -0
sky/utils/command_runner.py +3 -0
sky/utils/locks.py +5 -2
{skypilot_nightly-1.0.0.dev20251016.dist-info → skypilot_nightly-1.0.0.dev20251018.dist-info}/METADATA +277 -48
{skypilot_nightly-1.0.0.dev20251016.dist-info → skypilot_nightly-1.0.0.dev20251018.dist-info}/RECORD +40 -39
/sky/dashboard/out/_next/static/{pbgtEUoCUdmJyLHjgln5A → Rn37hj-nuHOYT-HwxSDXC}/_buildManifest.js +0 -0
/sky/dashboard/out/_next/static/{pbgtEUoCUdmJyLHjgln5A → Rn37hj-nuHOYT-HwxSDXC}/_ssgManifest.js +0 -0
{skypilot_nightly-1.0.0.dev20251016.dist-info → skypilot_nightly-1.0.0.dev20251018.dist-info}/WHEEL +0 -0
{skypilot_nightly-1.0.0.dev20251016.dist-info → skypilot_nightly-1.0.0.dev20251018.dist-info}/entry_points.txt +0 -0
{skypilot_nightly-1.0.0.dev20251016.dist-info → skypilot_nightly-1.0.0.dev20251018.dist-info}/licenses/LICENSE +0 -0
{skypilot_nightly-1.0.0.dev20251016.dist-info → skypilot_nightly-1.0.0.dev20251018.dist-info}/top_level.txt +0 -0

sky/dashboard/out/workspaces/[name].html CHANGED Viewed

	@@ -1 +1 @@
1	- <!DOCTYPE html><html><head><meta charSet="utf-8"/><meta name="viewport" content="width=device-width"/><meta name="next-head-count" content="2"/><link rel="preload" href="/dashboard/_next/static/css/4614e06482d7309e.css" as="style"/><link rel="stylesheet" href="/dashboard/_next/static/css/4614e06482d7309e.css" data-n-g=""/><noscript data-n-css=""></noscript><script defer="" nomodule="" src="/dashboard/_next/static/chunks/polyfills-78c92fac7aa8fdd8.js"></script><script src="/dashboard/_next/static/chunks/webpack-66f23594d38c7f16.js" defer=""></script><script src="/dashboard/_next/static/chunks/framework-cf60a09ccd051a10.js" defer=""></script><script src="/dashboard/_next/static/chunks/main-f15ccb73239a3bf1.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/_app-ce361c6959bc2001.js" defer=""></script><script src="/dashboard/_next/static/chunks/616-3d59f75e2ccf9321.js" defer=""></script><script src="/dashboard/_next/static/chunks/6130-2be46d70a38f1e82.js" defer=""></script><script src="/dashboard/_next/static/chunks/5739-d67458fcb1386c92.js" defer=""></script><script src="/dashboard/_next/static/chunks/7411-b15471acd2cba716.js" defer=""></script><script src="/dashboard/_next/static/chunks/1272-1ef0bf0237faccdb.js" defer=""></script><script src="/dashboard/_next/static/chunks/7359-c8d04e06886000b3.js" defer=""></script><script src="/dashboard/_next/static/chunks/6989-01359c57e018caa4.js" defer=""></script><script src="/dashboard/_next/static/chunks/3850-ff4a9a69d978632b.js" defer=""></script><script src="/dashboard/_next/static/chunks/8969-66237729cdf9749e.js" defer=""></script><script src="/dashboard/_next/static/chunks/6990-f6818c84ed8f1c86.js" defer=""></script><script src="/dashboard/_next/static/chunks/6135-4b4d5e824b7f9d3c.js" defer=""></script><script src="/dashboard/_next/static/chunks/1121-d0782b9251f0fcd3.js" defer=""></script><script src="/dashboard/_next/static/chunks/6601-06114c982db410b6.js" defer=""></script><script src="/dashboard/_next/static/chunks/3015-7e0e8f06bb2f881c.js" defer=""></script><script src="/dashboard/_next/static/chunks/1141-3b40c39626f99c89.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/workspaces/%5Bname%5D-e8688c35c06f0ac5.js" defer=""></script><script src="/dashboard/_next/static/~~pbgtEUoCUdmJyLHjgln5A~~/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/~~pbgtEUoCUdmJyLHjgln5A~~/_ssgManifest.js" defer=""></script></head><body><div id="__next"></div><script id="__NEXT_DATA__" type="application/json">{"props":{"pageProps":{}},"page":"/workspaces/[name]","query":{},"buildId":"~~pbgtEUoCUdmJyLHjgln5A~~","assetPrefix":"/dashboard","nextExport":true,"autoExport":true,"isFallback":false,"scriptLoader":[]}</script></body></html>
1	+ <!DOCTYPE html><html><head><meta charSet="utf-8"/><meta name="viewport" content="width=device-width"/><meta name="next-head-count" content="2"/><link rel="preload" href="/dashboard/_next/static/css/4614e06482d7309e.css" as="style"/><link rel="stylesheet" href="/dashboard/_next/static/css/4614e06482d7309e.css" data-n-g=""/><noscript data-n-css=""></noscript><script defer="" nomodule="" src="/dashboard/_next/static/chunks/polyfills-78c92fac7aa8fdd8.js"></script><script src="/dashboard/_next/static/chunks/webpack-66f23594d38c7f16.js" defer=""></script><script src="/dashboard/_next/static/chunks/framework-cf60a09ccd051a10.js" defer=""></script><script src="/dashboard/_next/static/chunks/main-f15ccb73239a3bf1.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/_app-ce361c6959bc2001.js" defer=""></script><script src="/dashboard/_next/static/chunks/616-3d59f75e2ccf9321.js" defer=""></script><script src="/dashboard/_next/static/chunks/6130-2be46d70a38f1e82.js" defer=""></script><script src="/dashboard/_next/static/chunks/5739-d67458fcb1386c92.js" defer=""></script><script src="/dashboard/_next/static/chunks/7411-b15471acd2cba716.js" defer=""></script><script src="/dashboard/_next/static/chunks/1272-1ef0bf0237faccdb.js" defer=""></script><script src="/dashboard/_next/static/chunks/7359-c8d04e06886000b3.js" defer=""></script><script src="/dashboard/_next/static/chunks/6989-01359c57e018caa4.js" defer=""></script><script src="/dashboard/_next/static/chunks/3850-ff4a9a69d978632b.js" defer=""></script><script src="/dashboard/_next/static/chunks/8969-66237729cdf9749e.js" defer=""></script><script src="/dashboard/_next/static/chunks/6990-f6818c84ed8f1c86.js" defer=""></script><script src="/dashboard/_next/static/chunks/6135-4b4d5e824b7f9d3c.js" defer=""></script><script src="/dashboard/_next/static/chunks/1121-d0782b9251f0fcd3.js" defer=""></script><script src="/dashboard/_next/static/chunks/6601-06114c982db410b6.js" defer=""></script><script src="/dashboard/_next/static/chunks/3015-7e0e8f06bb2f881c.js" defer=""></script><script src="/dashboard/_next/static/chunks/1141-3b40c39626f99c89.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/workspaces/%5Bname%5D-e8688c35c06f0ac5.js" defer=""></script><script src="/dashboard/_next/static/Rn37hj-nuHOYT-HwxSDXC/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/Rn37hj-nuHOYT-HwxSDXC/_ssgManifest.js" defer=""></script></head><body><div id="__next"></div><script id="__NEXT_DATA__" type="application/json">{"props":{"pageProps":{}},"page":"/workspaces/[name]","query":{},"buildId":"Rn37hj-nuHOYT-HwxSDXC","assetPrefix":"/dashboard","nextExport":true,"autoExport":true,"isFallback":false,"scriptLoader":[]}</script></body></html>

sky/dashboard/out/workspaces.html CHANGED Viewed

	@@ -1 +1 @@
1	- <!DOCTYPE html><html><head><meta charSet="utf-8"/><meta name="viewport" content="width=device-width"/><meta name="next-head-count" content="2"/><link rel="preload" href="/dashboard/_next/static/css/4614e06482d7309e.css" as="style"/><link rel="stylesheet" href="/dashboard/_next/static/css/4614e06482d7309e.css" data-n-g=""/><noscript data-n-css=""></noscript><script defer="" nomodule="" src="/dashboard/_next/static/chunks/polyfills-78c92fac7aa8fdd8.js"></script><script src="/dashboard/_next/static/chunks/webpack-66f23594d38c7f16.js" defer=""></script><script src="/dashboard/_next/static/chunks/framework-cf60a09ccd051a10.js" defer=""></script><script src="/dashboard/_next/static/chunks/main-f15ccb73239a3bf1.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/_app-ce361c6959bc2001.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/workspaces-69c80d677d3c2949.js" defer=""></script><script src="/dashboard/_next/static/~~pbgtEUoCUdmJyLHjgln5A~~/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/~~pbgtEUoCUdmJyLHjgln5A~~/_ssgManifest.js" defer=""></script></head><body><div id="__next"></div><script id="__NEXT_DATA__" type="application/json">{"props":{"pageProps":{}},"page":"/workspaces","query":{},"buildId":"~~pbgtEUoCUdmJyLHjgln5A~~","assetPrefix":"/dashboard","nextExport":true,"autoExport":true,"isFallback":false,"scriptLoader":[]}</script></body></html>
1	+ <!DOCTYPE html><html><head><meta charSet="utf-8"/><meta name="viewport" content="width=device-width"/><meta name="next-head-count" content="2"/><link rel="preload" href="/dashboard/_next/static/css/4614e06482d7309e.css" as="style"/><link rel="stylesheet" href="/dashboard/_next/static/css/4614e06482d7309e.css" data-n-g=""/><noscript data-n-css=""></noscript><script defer="" nomodule="" src="/dashboard/_next/static/chunks/polyfills-78c92fac7aa8fdd8.js"></script><script src="/dashboard/_next/static/chunks/webpack-66f23594d38c7f16.js" defer=""></script><script src="/dashboard/_next/static/chunks/framework-cf60a09ccd051a10.js" defer=""></script><script src="/dashboard/_next/static/chunks/main-f15ccb73239a3bf1.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/_app-ce361c6959bc2001.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/workspaces-69c80d677d3c2949.js" defer=""></script><script src="/dashboard/_next/static/Rn37hj-nuHOYT-HwxSDXC/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/Rn37hj-nuHOYT-HwxSDXC/_ssgManifest.js" defer=""></script></head><body><div id="__next"></div><script id="__NEXT_DATA__" type="application/json">{"props":{"pageProps":{}},"page":"/workspaces","query":{},"buildId":"Rn37hj-nuHOYT-HwxSDXC","assetPrefix":"/dashboard","nextExport":true,"autoExport":true,"isFallback":false,"scriptLoader":[]}</script></body></html>

sky/provision/fluidstack/instance.py CHANGED Viewed

@@ -3,11 +3,11 @@ import os
 import time
 from typing import Any, Dict, List, Optional, Tuple
-from sky import authentication as auth
 from sky import exceptions
 from sky import sky_logging
 from sky.provision import common
 from sky.provision.fluidstack import fluidstack_utils as utils
+from sky.utils import auth_utils
 from sky.utils import command_runner
 from sky.utils import common_utils
 from sky.utils import status_lib
@@ -27,7 +27,7 @@ logger = sky_logging.init_logger(__name__)
 def get_internal_ip(node_info: Dict[str, Any]) -> None:
     node_info['internal_ip'] = node_info['ip_address']
-    private_key_path, _ = auth.get_or_generate_keys()
+    private_key_path, _ = auth_utils.get_or_generate_keys()
     runner = command_runner.SSHCommandRunner(
         (node_info['ip_address'], 22),
         ssh_user='ubuntu',

sky/provision/seeweb/instance.py CHANGED Viewed

@@ -9,7 +9,6 @@ import subprocess
 import time
 from typing import Any, Dict, List, Optional, Tuple
-from sky import authentication as auth
 from sky import sky_logging
 from sky.adaptors import seeweb as seeweb_adaptor
 from sky.provision import common
@@ -17,6 +16,7 @@ from sky.provision.common import ClusterInfo
 from sky.provision.common import InstanceInfo
 from sky.provision.common import ProvisionConfig
 from sky.provision.common import ProvisionRecord
+from sky.utils import auth_utils
 from sky.utils import command_runner  # Unified SSH helper
 from sky.utils import common_utils
 from sky.utils import status_lib
@@ -75,7 +75,7 @@ class SeewebNodeProvider:
         if self.config and self.config.authentication_config:
             key_path = self.config.authentication_config.get('ssh_private_key')
         if not key_path:
-            key_path, _ = auth.get_or_generate_keys()
+            key_path, _ = auth_utils.get_or_generate_keys()
         return os.path.expanduser(key_path)
     # ------------------------------------------------------------------ #
@@ -661,7 +661,7 @@ def _ping_server_standalone(server_ip: str) -> bool:
 def _check_ssh_ready_standalone(server_ip: str) -> bool:
     """Check that SSH is available on the server (standalone version)."""
     try:
-        private_key_path, _ = auth.get_or_generate_keys()
+        private_key_path, _ = auth_utils.get_or_generate_keys()
         private_key_path = os.path.expanduser(private_key_path)
         ssh_user = 'ecuser'
         result = subprocess.run([

sky/serve/service.py CHANGED Viewed

@@ -13,7 +13,6 @@ from typing import Dict
 import filelock
-from sky import authentication
 from sky import exceptions
 from sky import global_user_state
 from sky import sky_logging
@@ -28,6 +27,7 @@ from sky.serve import replica_managers
 from sky.serve import serve_state
 from sky.serve import serve_utils
 from sky.skylet import constants as skylet_constants
+from sky.utils import auth_utils
 from sky.utils import common_utils
 from sky.utils import controller_utils
 from sky.utils import subprocess_utils
@@ -227,7 +227,7 @@ def _start(service_name: str, tmp_task_yaml: str, job_id: int, entrypoint: str):
     """
     # Generate ssh key pair to avoid race condition when multiple sky.launch
     # are executed at the same time.
-    authentication.get_or_generate_keys()
+    auth_utils.get_or_generate_keys()
     # Initialize database record for the service.
     task = task_lib.Task.from_yaml(tmp_task_yaml)

sky/server/requests/preconditions.py CHANGED Viewed

@@ -162,8 +162,8 @@ class ClusterStartCompletePrecondition(Precondition):
         requests = await api_requests.get_request_tasks_async(
             req_filter=api_requests.RequestTaskFilter(
                 status=[
-                    api_requests.RequestStatus.RUNNING,
-                    api_requests.RequestStatus.PENDING
+                    api_requests.RequestStatus.PENDING,
+                    api_requests.RequestStatus.RUNNING
                 ],
                 include_request_names=['sky.launch', 'sky.start'],
                 cluster_names=[self.cluster_name]))

sky/server/requests/requests.py CHANGED Viewed

@@ -398,9 +398,9 @@ def kill_cluster_requests(cluster_name: str, exclude_request_name: str):
     request_ids = [
         request_task.request_id
         for request_task in get_request_tasks(req_filter=RequestTaskFilter(
-            cluster_names=[cluster_name],
             status=[RequestStatus.PENDING, RequestStatus.RUNNING],
-            exclude_request_names=[exclude_request_name]))
+            exclude_request_names=[exclude_request_name],
+            cluster_names=[cluster_name]))
     ]
     kill_requests(request_ids)
@@ -422,10 +422,10 @@ def kill_requests(request_ids: Optional[List[str]] = None,
         request_ids = [
             request_task.request_id
             for request_task in get_request_tasks(req_filter=RequestTaskFilter(
-                user_id=user_id,
-                status=[RequestStatus.RUNNING, RequestStatus.PENDING],
+                status=[RequestStatus.PENDING, RequestStatus.RUNNING],
                 # Avoid cancelling the cancel request itself.
-                exclude_request_names=['sky.api_cancel']))
+                exclude_request_names=['sky.api_cancel'],
+                user_id=user_id))
         ]
     cancelled_request_ids = []
     for request_id in request_ids:
@@ -497,6 +497,21 @@ def create_table(cursor, conn):
     db_utils.add_column_to_table(cursor, conn, REQUEST_TABLE, COL_FINISHED_AT,
                                  'REAL')
+    # Add an index on (status, name) to speed up queries
+    # that filter on these columns.
+    cursor.execute(f"""\
+        CREATE INDEX IF NOT EXISTS status_name_idx ON {REQUEST_TABLE} (status, name) WHERE status IN ('PENDING', 'RUNNING');
+    """)
+    # Add an index on cluster_name to speed up queries
+    # that filter on this column.
+    cursor.execute(f"""\
+        CREATE INDEX IF NOT EXISTS cluster_name_idx ON {REQUEST_TABLE} ({COL_CLUSTER_NAME}) WHERE status IN ('PENDING', 'RUNNING');
+    """)
+    # Add an index on created_at to speed up queries that sort on this column.
+    cursor.execute(f"""\
+        CREATE INDEX IF NOT EXISTS created_at_idx ON {REQUEST_TABLE} (created_at);
+    """)
 _DB = None
 _init_db_lock = threading.Lock()
@@ -753,6 +768,10 @@ class RequestTaskFilter:
             status_list_str = ','.join(
                 repr(status.value) for status in self.status)
             filters.append(f'status IN ({status_list_str})')
+        if self.include_request_names is not None:
+            request_names_str = ','.join(
+                repr(name) for name in self.include_request_names)
+            filters.append(f'name IN ({request_names_str})')
         if self.exclude_request_names is not None:
             exclude_request_names_str = ','.join(
                 repr(name) for name in self.exclude_request_names)
@@ -764,10 +783,6 @@ class RequestTaskFilter:
         if self.user_id is not None:
             filters.append(f'{COL_USER_ID} = ?')
             filter_params.append(self.user_id)
-        if self.include_request_names is not None:
-            request_names_str = ','.join(
-                repr(name) for name in self.include_request_names)
-            filters.append(f'name IN ({request_names_str})')
         if self.finished_before is not None:
             filters.append('finished_at < ?')
             filter_params.append(self.finished_before)
@@ -800,6 +815,10 @@ def get_request_tasks(req_filter: RequestTaskFilter) -> List[Request]:
         rows = cursor.fetchall()
         if rows is None:
             return []
+    if req_filter.fields:
+        rows = [
+            _update_request_row_fields(row, req_filter.fields) for row in rows
+        ]
     return [Request.from_row(row) for row in rows]
@@ -812,21 +831,10 @@ async def get_request_tasks_async(
     async with _DB.execute_fetchall_async(*req_filter.build_query()) as rows:
         if not rows:
             return []
-    return [Request.from_row(row) for row in rows]
-@init_db_async
-@metrics_lib.time_me_async
-async def get_request_tasks_with_fields_async(
-    req_filter: RequestTaskFilter,
-    fields: Optional[List[str]] = None,
-) -> List[Request]:
-    """Async version of get_request_tasks."""
-    assert _DB is not None
-    async with _DB.execute_fetchall_async(*req_filter.build_query()) as rows:
-        if not rows:
-            return []
-    rows = [_update_request_row_fields(row, fields) for row in rows]
+    if req_filter.fields:
+        rows = [
+            _update_request_row_fields(row, req_filter.fields) for row in rows
+        ]
     return [Request.from_row(row) for row in rows]

sky/server/server.py CHANGED Viewed

@@ -1667,14 +1667,12 @@ async def api_status(
                 requests_lib.RequestStatus.PENDING,
                 requests_lib.RequestStatus.RUNNING,
             ]
-        request_tasks = await requests_lib.get_request_tasks_with_fields_async(
+        request_tasks = await requests_lib.get_request_tasks_async(
             req_filter=requests_lib.RequestTaskFilter(
                 status=statuses,
                 limit=limit,
                 fields=fields,
-            ),
-            fields=fields,
-        )
+            ))
         return requests_lib.encode_requests(request_tasks)
     else:
         encoded_request_tasks = []
@@ -2058,7 +2056,8 @@ if __name__ == '__main__':
         uvicorn_config = uvicorn.Config('sky.server.server:app',
                                         host=cmd_args.host,
                                         port=cmd_args.port,
-                                        workers=num_workers)
+                                        workers=num_workers,
+                                        ws_per_message_deflate=False)
         skyuvicorn.run(uvicorn_config,
                        max_db_connections=config.num_db_connections_per_worker)
     except Exception as exc:  # pylint: disable=broad-except

sky/server/stream_utils.py CHANGED Viewed

@@ -215,11 +215,18 @@ async def _tail_log_file(
             # periodically to see if provisioning is done.
             if cluster_name is not None and should_check_status:
                 last_status_check_time = current_time
-                cluster_record = await (
+                cluster_status = await (
                     global_user_state.get_status_from_cluster_name_async(
                         cluster_name))
-                if (cluster_record is None or
-                        cluster_record != status_lib.ClusterStatus.INIT):
+                if cluster_status is None:
+                    logger.debug(
+                        'Stop tailing provision logs for cluster'
+                        f' status for cluster {cluster_name} not found')
+                    break
+                if cluster_status != status_lib.ClusterStatus.INIT:
+                    logger.debug(f'Stop tailing provision logs for cluster'
+                                 f' {cluster_name} has status {cluster_status} '
+                                 '(not in INIT state)')
                     break
             if current_time - last_heartbeat_time >= _HEARTBEAT_INTERVAL:
                 # Currently just used to keep the connection busy, refer to

sky/setup_files/dependencies.py CHANGED Viewed

@@ -86,7 +86,6 @@ install_requires = [
     'types-paramiko',
     'alembic',
     'aiohttp',
-    'aiosqlite',
     'anyio',
 ]
@@ -104,6 +103,10 @@ GRPC = 'grpcio>=1.63.0'
 PROTOBUF = 'protobuf>=5.26.1, < 7.0.0'
 server_dependencies = [
+    # TODO: Some of these dependencies are also specified in install_requires,
+    # so they are redundant here. We should figure out if they are only needed
+    # on the server (should remove from install_requires), or if they are needed
+    # on the client (should remove from here).
     'casbin',
     'sqlalchemy_adapter',
     'passlib',
@@ -148,7 +151,7 @@ aws_dependencies = [
 # a few places.
 AZURE_CLI = 'azure-cli>=2.65.0'
-extras_require: Dict[str, List[str]] = {
+cloud_dependencies: Dict[str, List[str]] = {
     'aws': aws_dependencies,
     # TODO(zongheng): azure-cli is huge and takes a long time to install.
     # Tracked in: https://github.com/Azure/azure-cli/issues/7387
@@ -191,7 +194,6 @@ extras_require: Dict[str, List[str]] = {
         'kubernetes>=20.0.0,!=32.0.0', 'websockets', 'python-dateutil'
     ],
     'ssh': ['kubernetes>=20.0.0,!=32.0.0', 'websockets', 'python-dateutil'],
-    'remote': remote,
     # For the container registry auth api. Reference:
     # https://github.com/runpod/runpod-python/releases/tag/1.6.1
     # RunPod needs a TOML parser to read ~/.runpod/config.toml. On Python 3.11+
@@ -221,13 +223,11 @@ extras_require: Dict[str, List[str]] = {
     ] + aws_dependencies,
     'hyperbolic': [],  # No dependencies needed for hyperbolic
     'seeweb': ['ecsapi>=0.2.0'],
-    'server': server_dependencies,
     'shadeform': [],  # No dependencies needed for shadeform
 }
 # Calculate which clouds should be included in the [all] installation.
-clouds_for_all = set(extras_require)
-clouds_for_all.remove('remote')
+clouds_for_all = set(cloud_dependencies)
 if sys.version_info < (3, 10):
     # Nebius needs python3.10. If python 3.9 [all] will not install nebius
@@ -242,5 +242,16 @@ if sys.version_info >= (3, 12):
     # TODO: Remove once https://github.com/vast-ai/vast-sdk/pull/6 is released
     clouds_for_all.remove('vast')
-extras_require['all'] = list(
-    set().union(*[extras_require[cloud] for cloud in clouds_for_all]))
+cloud_extras = {
+    cloud: dependencies + server_dependencies
+    for cloud, dependencies in cloud_dependencies.items()
+}
+extras_require: Dict[str, List[str]] = {
+    # Include server_dependencies with each cloud.
+    **cloud_extras,
+    'all': list(set().union(*[cloud_extras[cloud] for cloud in clouds_for_all])
+               ),
+    'remote': remote,
+    'server': server_dependencies,
+}

sky/utils/auth_utils.py ADDED Viewed

@@ -0,0 +1,153 @@
+"""Utils for managing SkyPilot SSH key pairs."""
+import functools
+import os
+from typing import Tuple
+import filelock
+from sky import global_user_state
+from sky import sky_logging
+from sky.utils import common_utils
+logger = sky_logging.init_logger(__name__)
+MAX_TRIALS = 64
+# TODO(zhwu): Support user specified key pair.
+# We intentionally not have the ssh key pair to be stored in
+# ~/.sky/api_server/clients, i.e. sky.server.common.API_SERVER_CLIENT_DIR,
+# because ssh key pair need to persist across API server restarts, while
+# the former dir is empheral.
+_SSH_KEY_PATH_PREFIX = '~/.sky/clients/{user_hash}/ssh'
+def get_ssh_key_and_lock_path(user_hash: str) -> Tuple[str, str, str]:
+    user_ssh_key_prefix = _SSH_KEY_PATH_PREFIX.format(user_hash=user_hash)
+    os.makedirs(os.path.expanduser(user_ssh_key_prefix),
+                exist_ok=True,
+                mode=0o700)
+    private_key_path = os.path.join(user_ssh_key_prefix, 'sky-key')
+    public_key_path = os.path.join(user_ssh_key_prefix, 'sky-key.pub')
+    lock_path = os.path.join(user_ssh_key_prefix, '.__internal-sky-key.lock')
+    return private_key_path, public_key_path, lock_path
+def _generate_rsa_key_pair() -> Tuple[str, str]:
+    # Keep the import of the cryptography local to avoid expensive
+    # third-party imports when not needed.
+    # pylint: disable=import-outside-toplevel
+    from cryptography.hazmat.backends import default_backend
+    from cryptography.hazmat.primitives import serialization
+    from cryptography.hazmat.primitives.asymmetric import rsa
+    key = rsa.generate_private_key(backend=default_backend(),
+                                   public_exponent=65537,
+                                   key_size=2048)
+    private_key = key.private_bytes(
+        encoding=serialization.Encoding.PEM,
+        format=serialization.PrivateFormat.TraditionalOpenSSL,
+        encryption_algorithm=serialization.NoEncryption()).decode(
+            'utf-8').strip()
+    public_key = key.public_key().public_bytes(
+        serialization.Encoding.OpenSSH,
+        serialization.PublicFormat.OpenSSH).decode('utf-8').strip()
+    return public_key, private_key
+def _save_key_pair(private_key_path: str, public_key_path: str,
+                   private_key: str, public_key: str) -> None:
+    key_dir = os.path.dirname(private_key_path)
+    os.makedirs(key_dir, exist_ok=True, mode=0o700)
+    with open(
+            private_key_path,
+            'w',
+            encoding='utf-8',
+            opener=functools.partial(os.open, mode=0o600),
+    ) as f:
+        f.write(private_key)
+    with open(public_key_path,
+              'w',
+              encoding='utf-8',
+              opener=functools.partial(os.open, mode=0o644)) as f:
+        f.write(public_key)
+def get_or_generate_keys() -> Tuple[str, str]:
+    """Returns the absolute private and public key paths."""
+    user_hash = common_utils.get_user_hash()
+    private_key_path, public_key_path, lock_path = get_ssh_key_and_lock_path(
+        user_hash)
+    private_key_path = os.path.expanduser(private_key_path)
+    public_key_path = os.path.expanduser(public_key_path)
+    lock_path = os.path.expanduser(lock_path)
+    lock_dir = os.path.dirname(lock_path)
+    # We should have the folder ~/.sky/generated/ssh to have 0o700 permission,
+    # as the ssh configs will be written to this folder as well in
+    # backend_utils.SSHConfigHelper
+    os.makedirs(lock_dir, exist_ok=True, mode=0o700)
+    with filelock.FileLock(lock_path, timeout=10):
+        if not os.path.exists(private_key_path):
+            ssh_public_key, ssh_private_key, exists = (
+                global_user_state.get_ssh_keys(user_hash))
+            if not exists:
+                ssh_public_key, ssh_private_key = _generate_rsa_key_pair()
+                global_user_state.set_ssh_keys(user_hash, ssh_public_key,
+                                               ssh_private_key)
+            _save_key_pair(private_key_path, public_key_path, ssh_private_key,
+                           ssh_public_key)
+    assert os.path.exists(public_key_path), (
+        'Private key found, but associated public key '
+        f'{public_key_path} does not exist.')
+    return private_key_path, public_key_path
+def create_ssh_key_files_from_db(private_key_path: str) -> bool:
+    """Creates the ssh key files from the database.
+    Returns:
+        True if the ssh key files are created successfully, False otherwise.
+    """
+    # Assume private key path is in the format of
+    # ~/.sky/clients/<user_hash>/ssh/sky-key
+    separated_path = os.path.normpath(private_key_path).split(os.path.sep)
+    assert separated_path[-1] == 'sky-key'
+    assert separated_path[-2] == 'ssh'
+    user_hash = separated_path[-3]
+    private_key_path_generated, public_key_path, lock_path = (
+        get_ssh_key_and_lock_path(user_hash))
+    assert private_key_path == os.path.expanduser(private_key_path_generated), (
+        f'Private key path {private_key_path} does not '
+        'match the generated path '
+        f'{os.path.expanduser(private_key_path_generated)}')
+    private_key_path = os.path.expanduser(private_key_path)
+    public_key_path = os.path.expanduser(public_key_path)
+    lock_path = os.path.expanduser(lock_path)
+    lock_dir = os.path.dirname(lock_path)
+    if os.path.exists(private_key_path) and os.path.exists(public_key_path):
+        return True
+    # We should have the folder ~/.sky/generated/ssh to have 0o700 permission,
+    # as the ssh configs will be written to this folder as well in
+    # backend_utils.SSHConfigHelper
+    os.makedirs(lock_dir, exist_ok=True, mode=0o700)
+    with filelock.FileLock(lock_path, timeout=10):
+        if not os.path.exists(private_key_path):
+            ssh_public_key, ssh_private_key, exists = (
+                global_user_state.get_ssh_keys(user_hash))
+            if not exists:
+                logger.debug(f'SSH keys not found for user {user_hash}')
+                return False
+            _save_key_pair(private_key_path, public_key_path, ssh_private_key,
+                           ssh_public_key)
+    assert os.path.exists(public_key_path), (
+        'Private key found, but associated public key '
+        f'{public_key_path} does not exist.')
+    return True

sky/utils/command_runner.py CHANGED Viewed

@@ -14,6 +14,7 @@ from sky import exceptions
 from sky import sky_logging
 from sky.skylet import constants
 from sky.skylet import log_lib
+from sky.utils import auth_utils
 from sky.utils import common_utils
 from sky.utils import context_utils
 from sky.utils import control_master_utils
@@ -649,6 +650,8 @@ class SSHCommandRunner(CommandRunner):
         self.disable_control_master = (
             disable_control_master or
             control_master_utils.should_disable_control_master())
+        # ensure the ssh key files are created from the database
+        auth_utils.create_ssh_key_files_from_db(ssh_private_key)
         if docker_user is not None:
             assert port is None or port == 22, (
                 f'port must be None or 22 for docker_user, got {port}.')

sky/utils/locks.py CHANGED Viewed

@@ -312,8 +312,11 @@ class PostgresLock(DistributedLock):
                 else:
                     self._connection.close()
             except Exception as e:  # pylint: disable=broad-except
-                logger.debug(
-                    f'Failed to invalidate or close postgres connection: {e}')
+                if invalidate:
+                    logger.debug(
+                        f'Failed to invalidate postgres connection: {e}')
+                else:
+                    logger.debug(f'Failed to close postgres connection: {e}')
             self._connection = None
     def is_locked(self) -> bool:

skypilot-nightly 1.0.0.dev20251016__py3-none-any.whl → 1.0.0.dev20251018__py3-none-any.whl

Potentially problematic release.

skypilot-nightly 1.0.0.dev20251016py3-none-any.whl → 1.0.0.dev20251018py3-none-any.whl