skypilot-nightly 1.0.0.dev20250905__py3-none-any.whl → 1.0.0.dev20251203__py3-none-any.whl

sky/provision/vsphere/common/vim_utils.py

@@ -4,7 +4,7 @@
 import re
 import subprocess
 import time
-from typing import List
+from typing import List, Union
 
 from sky import sky_logging
 from sky.adaptors import vsphere as vsphere_adaptor
@@ -15,7 +15,8 @@ DISPLAY_CONTROLLER_CLASS_ID_PREFIXES = ['03']
 VMWARE_VIRTUAL_DISPLAY_CONTROLLER_IDS = ['0000:00:0f.0']
 
 
-def get_objs_by_names(content, vimtype: type, names: List[str]):
+def get_objs_by_names(content, vimtype: Union[type, List[type]],
+                      names: List[str]):
     """    Get the vsphere managed object associated with a given text name
     """
     # Create a set for the names for faster lookups

sky/provision/vsphere/instance.py

@@ -30,9 +30,10 @@ HEAD_NODE_VALUE = '1'
 WORKER_NODE_VALUE = '0'
 
 
-def run_instances(region: str, cluster_name: str,
+def run_instances(region: str, cluster_name: str, cluster_name_on_cloud: str,
                   config: common.ProvisionConfig) -> common.ProvisionRecord:
     """See sky/provision/__init__.py"""
+    del cluster_name  # unused
     logger.info('New provision of Vsphere: run_instances().')
 
     resumed_instance_ids: List[str] = []
@@ -40,7 +41,7 @@ def run_instances(region: str, cluster_name: str,
     vc_object = _get_vc_object(region)
     vc_object.connect()
 
-    exist_instances = _get_filtered_instance(vc_object, cluster_name,
+    exist_instances = _get_filtered_instance(vc_object, cluster_name_on_cloud,
                                              config.provider_config)
     head_instance_id = _get_head_instance_id(exist_instances)
 
@@ -89,8 +90,8 @@ def run_instances(region: str, cluster_name: str,
             config, region, vc_object)
         # TODO: update logic for multi-node creation
         for _ in range(to_start_num):
-            created_instance_uuid = _create_instances(cluster_name, config,
-                                                      region, vc_object,
+            created_instance_uuid = _create_instances(cluster_name_on_cloud,
+                                                      config, region, vc_object,
                                                       vsphere_cluster_name)
             created_instance_ids.append(created_instance_uuid)
         if head_instance_id is None:
@@ -104,7 +105,7 @@ def run_instances(region: str, cluster_name: str,
         provider_name='vsphere',
         region=region,
         zone=vsphere_cluster_name,
-        cluster_name=cluster_name,
+        cluster_name=cluster_name_on_cloud,
         head_instance_id=head_instance_id,
         resumed_instance_ids=resumed_instance_ids,
         created_instance_ids=created_instance_ids,
@@ -397,9 +398,10 @@ def query_instances(
     cluster_name_on_cloud: str,
     provider_config: Optional[Dict[str, Any]] = None,
     non_terminated_only: bool = True,
+    retry_if_missing: bool = False,
 ) -> Dict[str, Tuple[Optional['status_lib.ClusterStatus'], Optional[str]]]:
     """See sky/provision/__init__.py"""
-    del cluster_name  # unused
+    del cluster_name, retry_if_missing  # unused
     logger.info('New provision of Vsphere: query_instances().')
     assert provider_config is not None, cluster_name_on_cloud
     region = provider_config['region']

sky/provision/vsphere/vsphere_utils.py

@@ -262,6 +262,10 @@ class VsphereClient:
 
     def get_pbm_manager(self):
         self.connect()
+        if self.servicemanager is None:
+            raise VsphereError('Failed to connect to vSphere.')
+        if self.servicemanager.si is None:
+            raise VsphereError('Failed to connect to vSphere.')
         pbm_si, pm_content = self._create_pbm_connection(  # pylint: disable=unused-variable
             self.servicemanager.si._stub)  # pylint: disable=protected-access
         pm = pm_content.profileManager
@@ -360,6 +364,8 @@ def initialize_vsphere_data():
         vcenter_name = vcenter['name']
         vc_object.connect()
         vc_servicemanager = vc_object.servicemanager
+        if vc_servicemanager is None or vc_servicemanager.content is None:
+            raise VsphereError('Failed to connect to vSphere.')
         vc_content = vc_servicemanager.content
 
         cluster_name_dicts = vc_object.clusters
@@ -370,4 +376,5 @@ def initialize_vsphere_data():
         initialize_images_csv(images_csv_path, vc_object, vcenter_name)
         initialize_instance_image_mapping_csv(vms_csv_path, images_csv_path,
                                               instance_image_mapping_csv_path)
-        vc_object.servicemanager.disconnect()
+        if vc_object.servicemanager is not None:
+            vc_object.servicemanager.disconnect()

sky/resources.py

@@ -1104,7 +1104,7 @@ class Resources:
         regions = self.cloud.regions_with_offering(self._instance_type,
                                                    self.accelerators,
                                                    self._use_spot, self._region,
-                                                   self._zone)
+                                                   self._zone, self)
         if self._image_id is not None and None not in self._image_id:
             regions = [r for r in regions if r.name in self._image_id]
 
@@ -1331,10 +1331,18 @@ class Resources:
                     clouds.CloudImplementationFeatures.IMAGE_ID
                 })
         except exceptions.NotSupportedError as e:
+            # Provide a more helpful error message for Lambda cloud
+            if self.cloud.is_same_cloud(clouds.Lambda()):
+                with ux_utils.print_exception_no_traceback():
+                    raise ValueError(
+                        'Lambda cloud only supports Docker images. '
+                        'Please prefix your image with "docker:" '
+                        '(e.g., image_id: docker:your-image-name).') from e
             with ux_utils.print_exception_no_traceback():
                 raise ValueError(
                     'image_id is only supported for AWS/GCP/Azure/IBM/OCI/'
-                    'Kubernetes, please explicitly specify the cloud.') from e
+                    'Kubernetes. For Lambda cloud, use "docker:" prefix for '
+                    'Docker images.') from e
 
         if self._region is not None:
             # If the image_id has None as key (region-agnostic),
@@ -1516,7 +1524,7 @@ class Resources:
         if self.accelerators is not None:
             hourly_cost += self.cloud.accelerators_to_hourly_cost(
                 self.accelerators, self.use_spot, self._region, self._zone)
-        return hourly_cost * hours
+        return float(hourly_cost * hours)
 
     def get_accelerators_str(self) -> str:
         accelerators = self.accelerators

sky/schemas/api/responses.py

@@ -5,8 +5,11 @@ from typing import Any, Dict, List, Optional
 
 import pydantic
 
+from sky import data
 from sky import models
+from sky.jobs import state as job_state
 from sky.server import common
+from sky.skylet import job_lib
 from sky.utils import status_lib
 
 
@@ -74,8 +77,13 @@ class APIHealthResponse(ResponseBaseModel):
     version: str = ''
     version_on_disk: str = ''
     commit: str = ''
+    # Whether basic auth on api server is enabled
     basic_auth_enabled: bool = False
     user: Optional[models.User] = None
+    # Whether service account token is enabled
+    service_account_token_enabled: bool = False
+    # Whether basic auth on ingress is enabled
+    ingress_basic_auth_enabled: bool = False
 
 
 class StatusResponse(ResponseBaseModel):
@@ -86,8 +94,8 @@ class StatusResponse(ResponseBaseModel):
     # backends.ResourceHandle, so we use Any here.
     # This is an internally facing field anyway, so it's less
     # of a problem that it's not typed.
-    handle: Any
-    last_use: str
+    handle: Optional[Any] = None
+    last_use: Optional[str] = None
     status: status_lib.ClusterStatus
     autostop: int
     to_down: bool
@@ -95,11 +103,8 @@ class StatusResponse(ResponseBaseModel):
     # metadata is a JSON, so we use Any here.
     metadata: Optional[Dict[str, Any]] = None
     cluster_hash: str
-    # pydantic cannot generate the pydantic-core schema for
-    # storage_mounts_metadata, so we use Any here.
-    storage_mounts_metadata: Optional[Dict[str, Any]] = None
     cluster_ever_up: bool
-    status_updated_at: int
+    status_updated_at: Optional[int] = None
     user_hash: str
     user_name: str
     config_hash: Optional[str] = None
@@ -118,9 +123,105 @@ class StatusResponse(ResponseBaseModel):
     cpus: Optional[str] = None
     memory: Optional[str] = None
     accelerators: Optional[str] = None
+    cluster_name_on_cloud: Optional[str] = None
+
+
+class ClusterJobRecord(ResponseBaseModel):
+    """Response for the cluster job queue endpoint."""
+    job_id: int
+    job_name: str
+    username: str
+    user_hash: str
+    submitted_at: float
+    # None if the job has not started yet.
+    start_at: Optional[float] = None
+    # None if the job has not ended yet.
+    end_at: Optional[float] = None
+    resources: str
+    status: job_lib.JobStatus
+    log_path: str
+    metadata: Dict[str, Any] = {}
 
 
 class UploadStatus(enum.Enum):
     """Status of the upload."""
     UPLOADING = 'uploading'
     COMPLETED = 'completed'
+
+
+class StorageRecord(ResponseBaseModel):
+    """Response for the storage list endpoint."""
+    name: str
+    launched_at: int
+    store: List[data.StoreType]
+    last_use: str
+    status: status_lib.StorageStatus
+
+
+# TODO (syang) figure out which fields are always present
+# and therefore can be non-optional.
+class ManagedJobRecord(ResponseBaseModel):
+    """A single managed job record."""
+    # The job_id in the spot table
+    task_job_id: Optional[int] = pydantic.Field(None, alias='_job_id')
+    job_id: Optional[int] = None
+    task_id: Optional[int] = None
+    job_name: Optional[str] = None
+    task_name: Optional[str] = None
+    job_duration: Optional[float] = None
+    workspace: Optional[str] = None
+    status: Optional[job_state.ManagedJobStatus] = None
+    schedule_state: Optional[str] = None
+    resources: Optional[str] = None
+    cluster_resources: Optional[str] = None
+    cluster_resources_full: Optional[str] = None
+    cloud: Optional[str] = None
+    region: Optional[str] = None
+    zone: Optional[str] = None
+    infra: Optional[str] = None
+    recovery_count: Optional[int] = None
+    details: Optional[str] = None
+    failure_reason: Optional[str] = None
+    user_name: Optional[str] = None
+    user_hash: Optional[str] = None
+    submitted_at: Optional[float] = None
+    start_at: Optional[float] = None
+    end_at: Optional[float] = None
+    user_yaml: Optional[str] = None
+    entrypoint: Optional[str] = None
+    metadata: Optional[Dict[str, Any]] = None
+    controller_pid: Optional[int] = None
+    controller_pid_started_at: Optional[float] = None
+    dag_yaml_path: Optional[str] = None
+    env_file_path: Optional[str] = None
+    last_recovered_at: Optional[float] = None
+    run_timestamp: Optional[str] = None
+    priority: Optional[int] = None
+    original_user_yaml_path: Optional[str] = None
+    pool: Optional[str] = None
+    pool_hash: Optional[str] = None
+    current_cluster_name: Optional[str] = None
+    job_id_on_pool_cluster: Optional[int] = None
+    accelerators: Optional[Dict[str, int]] = None
+
+
+class VolumeRecord(ResponseBaseModel):
+    """A single volume record."""
+    name: str
+    type: str
+    launched_at: int
+    cloud: str
+    region: Optional[str] = None
+    zone: Optional[str] = None
+    size: Optional[str] = None
+    config: Dict[str, Any]
+    name_on_cloud: str
+    user_hash: str
+    user_name: str
+    workspace: str
+    last_attached_at: Optional[int] = None
+    last_use: Optional[str] = None
+    status: Optional[str] = None
+    usedby_pods: List[str]
+    usedby_clusters: List[str]
+    is_ephemeral: bool = False

sky/schemas/db/global_user_state/008_skylet_ssh_tunnel_metadata.py

@@ -0,0 +1,34 @@
+"""Add skylet_ssh_tunnel_metadata to clusters.
+
+Revision ID: 008
+Revises: 007
+Create Date: 2025-09-09
+
+"""
+# pylint: disable=invalid-name
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+from sky.utils.db import db_utils
+
+# revision identifiers, used by Alembic.
+revision: str = '008'
+down_revision: Union[str, Sequence[str], None] = '007'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade():
+    """Add skylet_ssh_tunnel_metadata column to clusters."""
+    with op.get_context().autocommit_block():
+        db_utils.add_column_to_table_alembic('clusters',
+                                             'skylet_ssh_tunnel_metadata',
+                                             sa.LargeBinary(),
+                                             server_default=None)
+
+
+def downgrade():
+    """No-op for backward compatibility."""
+    pass

sky/schemas/db/global_user_state/009_last_activity_and_launched_at.py

@@ -0,0 +1,89 @@
+"""Add last_activity_time and launched_at to cluster history.
+
+Revision ID: 009
+Revises: 008
+Create Date: 2025-09-24
+
+"""
+# pylint: disable=invalid-name
+import pickle
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+from sky.utils.db import db_utils
+
+# revision identifiers, used by Alembic.
+revision: str = '009'
+down_revision: Union[str, Sequence[str], None] = '008'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade():
+    """Add last_activity_time and launched_at columns to cluster history."""
+    with op.get_context().autocommit_block():
+        # Add the columns with indices
+        db_utils.add_column_to_table_alembic('cluster_history',
+                                             'last_activity_time',
+                                             sa.Integer(),
+                                             server_default=None,
+                                             index=True)
+
+        db_utils.add_column_to_table_alembic('cluster_history',
+                                             'launched_at',
+                                             sa.Integer(),
+                                             server_default=None,
+                                             index=True)
+
+        # Populate the columns for existing rows
+        _populate_cluster_history_columns()
+
+
+def _populate_cluster_history_columns():
+    """Populate last_activity_time and launched_at for existing rows using
+    usage_intervals logic."""
+    connection = op.get_bind()
+
+    # Get all existing rows with usage_intervals
+    result = connection.execute(
+        sa.text('SELECT cluster_hash, usage_intervals FROM cluster_history '
+                'WHERE usage_intervals IS NOT NULL'))
+
+    for row in result:
+        cluster_hash = row[0]
+        usage_intervals_blob = row[1]
+
+        try:
+            # Deserialize the usage_intervals
+            usage_intervals = pickle.loads(usage_intervals_blob)
+
+            if usage_intervals:
+                # Calculate last_activity_time: end time of last interval
+                # or start time if still running
+                last_interval = usage_intervals[-1]
+                last_activity_time = (last_interval[1] if last_interval[1]
+                                      is not None else last_interval[0])
+
+                # Calculate launched_at: start time of first interval
+                launched_at = usage_intervals[0][0]
+
+                # Update the row with both calculated values
+                connection.execute(
+                    sa.text('UPDATE cluster_history '
+                            'SET last_activity_time = :last_activity_time, '
+                            'launched_at = :launched_at '
+                            'WHERE cluster_hash = :cluster_hash'), {
+                                'last_activity_time': last_activity_time,
+                                'launched_at': launched_at,
+                                'cluster_hash': cluster_hash
+                            })
+        except (pickle.PickleError, AttributeError, IndexError):
+            # Skip rows with corrupted or invalid usage_intervals
+            continue
+
+
+def downgrade():
+    """No-op for backward compatibility."""
+    pass

sky/schemas/db/global_user_state/010_save_ssh_key.py

@@ -0,0 +1,66 @@
+"""Add ssh keys in filesystem to global user state.
+
+Revision ID: 010
+Revises: 009
+Create Date: 2025-10-07
+
+"""
+import glob
+# pylint: disable=invalid-name
+import os
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+# revision identifiers, used by Alembic.
+revision: str = '010'
+down_revision: Union[str, Sequence[str], None] = '009'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade():
+    """Add last_activity_time and launched_at columns to cluster history."""
+    connection = op.get_bind()
+
+    match_dirs = glob.glob(os.path.expanduser('~/.sky/clients/*/ssh'))
+    file_user_hashes = set()
+    for match_dir in match_dirs:
+        user_hash = match_dir.split('/')[-2]
+        file_user_hashes.add(user_hash)
+
+    # Get all existing ssh keys
+    existing_user_hashes = set()
+    result = connection.execute(sa.text('SELECT user_hash FROM ssh_key'))
+    for row in result:
+        existing_user_hashes.add(row[0])
+
+    user_hashes_to_add = file_user_hashes - existing_user_hashes
+    for user_hash in user_hashes_to_add:
+        match_dir = os.path.join(os.path.expanduser('~/.sky/clients'),
+                                 user_hash, 'ssh')
+        public_key_path = os.path.join(match_dir, 'sky-key.pub')
+        private_key_path = os.path.join(match_dir, 'sky-key')
+        try:
+            with open(public_key_path, 'r', encoding='utf-8') as f:
+                public_key = f.read().strip()
+            with open(private_key_path, 'r', encoding='utf-8') as f:
+                private_key = f.read().strip()
+        except FileNotFoundError:
+            # Skip if the key files are not found
+            continue
+        connection.execute(
+            sa.text('INSERT INTO ssh_key '
+                    '(user_hash, ssh_public_key, ssh_private_key) '
+                    'VALUES (:user_hash, :ssh_public_key, :ssh_private_key) '
+                    'ON CONFLICT DO NOTHING'), {
+                        'user_hash': user_hash,
+                        'ssh_public_key': public_key,
+                        'ssh_private_key': private_key
+                    })
+
+
+def downgrade():
+    """No-op for backward compatibility."""
+    pass

sky/schemas/db/global_user_state/011_is_ephemeral.py

@@ -0,0 +1,34 @@
+"""Columns for whether the volume is ephemeral.
+
+Revision ID: 011
+Revises: 010
+Create Date: 2025-11-14
+
+"""
+# pylint: disable=invalid-name
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+from sky.utils.db import db_utils
+
+# revision identifiers, used by Alembic.π
+revision: str = '011'
+down_revision: Union[str, Sequence[str], None] = '010'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade():
+    """Add columns for whether the volume is ephemeral."""
+    with op.get_context().autocommit_block():
+        db_utils.add_column_to_table_alembic('volumes',
+                                             'is_ephemeral',
+                                             sa.Integer(),
+                                             server_default='0')
+
+
+def downgrade():
+    """Remove columns for whether the volume is ephemeral."""
+    pass

sky/schemas/db/kv_cache/001_initial_schema.py

@@ -0,0 +1,29 @@
+"""Initial schema for KV cache database
+
+Revision ID: 001
+Revises:
+Create Date: 2025-11-13 12:00:00.000000
+
+"""
+# pylint: disable=invalid-name
+from alembic import op
+
+from sky.utils.db import db_utils
+from sky.utils.db.kv_cache import Base
+
+# revision identifiers, used by Alembic.
+revision = '001'
+down_revision = None
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    with op.get_context().autocommit_block():
+        # Create any missing tables with current schema first
+        db_utils.add_all_tables_to_db_sqlalchemy(Base.metadata, op.get_bind())
+
+
+def downgrade():
+    # Drop all tables
+    Base.metadata.drop_all(bind=op.get_bind())

sky/schemas/db/serve_state/002_yaml_content.py

@@ -0,0 +1,34 @@
+"""Columns for cluster pool.
+
+Revision ID: 002
+Revises: 001
+Create Date: 2025-07-18
+
+"""
+# pylint: disable=invalid-name
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+from sky.utils.db import db_utils
+
+# revision identifiers, used by Alembic.
+revision: str = '002'
+down_revision: Union[str, Sequence[str], None] = '001'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade():
+    """Add columns for cluster pool."""
+    with op.get_context().autocommit_block():
+        db_utils.add_column_to_table_alembic('version_specs',
+                                             'yaml_content',
+                                             sa.Text(),
+                                             server_default=None)
+
+
+def downgrade():
+    """Remove columns for cluster pool."""
+    pass

sky/schemas/db/skypilot_config/001_initial_schema.py

@@ -0,0 +1,30 @@
+"""Initial schema for sky config database
+
+Revision ID: 001
+Revises:
+Create Date: 2025-10-21
+
+"""
+# pylint: disable=invalid-name
+from alembic import op
+
+from sky.skypilot_config import Base
+from sky.utils.db import db_utils
+
+# revision identifiers, used by Alembic.
+revision = '001'
+down_revision = None
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    """Create initial schema for config_yaml table"""
+    with op.get_context().autocommit_block():
+        # Create all tables with their current schema
+        db_utils.add_all_tables_to_db_sqlalchemy(Base.metadata, op.get_bind())
+
+
+def downgrade():
+    """Drop all tables"""
+    Base.metadata.drop_all(bind=op.get_bind())

sky/schemas/db/spot_jobs/002_cluster_pool.py

@@ -1,4 +1,4 @@
-"""Columns for cluster pool.
+"""Columns for pool.
 
 Revision ID: 002
 Revises: 001
@@ -21,7 +21,7 @@ depends_on: Union[str, Sequence[str], None] = None
 
 
 def upgrade():
-    """Add columns for cluster pool."""
+    """Add columns for pool."""
     with op.get_context().autocommit_block():
         db_utils.add_column_to_table_alembic('job_info',
                                              'pool',
@@ -38,5 +38,5 @@ def upgrade():
 
 
 def downgrade():
-    """Remove columns for cluster pool."""
+    """Remove columns for pool."""
     pass

sky/schemas/db/spot_jobs/004_job_file_contents.py

@@ -0,0 +1,42 @@
+"""Add columns for stored DAG/env file contents.
+
+Revision ID: 004
+Revises: 003
+Create Date: 2025-10-27
+
+"""
+# pylint: disable=invalid-name
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+from sky.utils.db import db_utils
+
+# revision identifiers, used by Alembic.
+revision: str = '004'
+down_revision: Union[str, Sequence[str], None] = '003'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade():
+    """Add columns to persist job file contents in the database."""
+    with op.get_context().autocommit_block():
+        db_utils.add_column_to_table_alembic('job_info',
+                                             'dag_yaml_content',
+                                             sa.Text(),
+                                             server_default=None)
+        db_utils.add_column_to_table_alembic('job_info',
+                                             'original_user_yaml_content',
+                                             sa.Text(),
+                                             server_default=None)
+        db_utils.add_column_to_table_alembic('job_info',
+                                             'env_file_content',
+                                             sa.Text(),
+                                             server_default=None)
+
+
+def downgrade():
+    """No downgrade logic."""
+    pass