skypilot-nightly 1.0.0.dev20250814__py3-none-any.whl → 1.0.0.dev20250816__py3-none-any.whl

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 = '58649973a7c706775528a419f46ae024e59f4603'
+_SKYPILOT_COMMIT_SHA = 'bff0c2a2d33d0990092c7c33a532359ffe1b6c56'
 
 
 def _get_git_commit():
@@ -35,7 +35,7 @@ def _get_git_commit():
 
 
 __commit__ = _get_git_commit()
-__version__ = '1.0.0.dev20250814'
+__version__ = '1.0.0.dev20250816'
 __root_dir__ = os.path.dirname(os.path.abspath(__file__))
 
 

sky/adaptors/nebius.py

@@ -1,7 +1,8 @@
 """Nebius cloud adaptor."""
+import asyncio
 import os
 import threading
-from typing import List, Optional
+from typing import Any, Awaitable, List, Optional
 
 from sky import sky_logging
 from sky import skypilot_config
@@ -9,8 +10,49 @@ from sky.adaptors import common
 from sky.utils import annotations
 from sky.utils import ux_utils
 
+# Default read timeout for nebius SDK
+READ_TIMEOUT = 10
+
 logger = sky_logging.init_logger(__name__)
 
+_loop_lock = threading.Lock()
+_loop = None
+
+
+def _get_event_loop() -> asyncio.AbstractEventLoop:
+    """Get event loop for nebius sdk."""
+    global _loop
+
+    if _loop is not None:
+        return _loop
+
+    with _loop_lock:
+        if _loop is None:
+            # Create a new event loop in a dedicated thread
+            _loop = asyncio.new_event_loop()
+            threading.Thread(target=_loop.run_forever, daemon=True).start()
+
+        return _loop
+
+
+def sync_call(awaitable: Awaitable[Any]) -> Any:
+    """Synchronously run an awaitable in coroutine.
+
+    This wrapper is used to workaround:
+    https://github.com/nebius/pysdk/issues/76
+
+    Uses a dedicated background event loop to avoid conflicts
+    with existing asyncio contexts and prevent BlockingIOError.
+    """
+    loop = _get_event_loop()
+    future = asyncio.run_coroutine_threadsafe(_coro(awaitable), loop)
+    return future.result()
+
+
+async def _coro(awaitable: Awaitable[Any]) -> Any:
+    """Wrapper coroutine for awaitable."""
+    return await awaitable
+
 
 def tenant_id_path() -> str:
     return '~/.nebius/NEBIUS_TENANT_ID.txt'

sky/backends/backend.py

@@ -147,8 +147,9 @@ class Backend(Generic[_ResourceHandleType]):
     def teardown(self,
                  handle: _ResourceHandleType,
                  terminate: bool,
-                 purge: bool = False) -> None:
-        self._teardown(handle, terminate, purge)
+                 purge: bool = False,
+                 explicitly_requested: bool = False) -> None:
+        self._teardown(handle, terminate, purge, explicitly_requested)
 
     def register_info(self, **kwargs) -> None:
         """Register backend-specific information."""
@@ -200,5 +201,6 @@ class Backend(Generic[_ResourceHandleType]):
     def _teardown(self,
                   handle: _ResourceHandleType,
                   terminate: bool,
-                  purge: bool = False):
+                  purge: bool = False,
+                  explicitly_requested: bool = False):
         raise NotImplementedError

sky/backends/backend_utils.py

@@ -2017,7 +2017,15 @@ def _update_cluster_status(cluster_name: str) -> Optional[Dict[str, Any]]:
     if handle.cluster_yaml is None:
         # Remove cluster from db since this cluster does not have a config file
         # or any other ongoing requests
-        global_user_state.remove_cluster(cluster_name, terminate=True)
+        global_user_state.add_cluster_event(
+            cluster_name,
+            None,
+            'Cluster has no YAML file. Removing the cluster from cache.',
+            global_user_state.ClusterEventType.STATUS_CHANGE,
+            nop_if_duplicate=True)
+        global_user_state.remove_cluster(cluster_name,
+                                         terminate=True,
+                                         remove_events=True)
         logger.debug(f'Cluster {cluster_name!r} has no YAML file. '
                      'Removing the cluster from cache.')
         return None
@@ -2137,7 +2145,7 @@ def _update_cluster_status(cluster_name: str) -> Optional[Dict[str, Any]]:
         global_user_state.add_cluster_event(
             cluster_name,
             status_lib.ClusterStatus.UP,
-            'All nodes up + ray cluster healthy.',
+            'All nodes up; SkyPilot runtime healthy.',
             global_user_state.ClusterEventType.STATUS_CHANGE,
             nop_if_duplicate=True)
         global_user_state.add_or_update_cluster(cluster_name,
@@ -2277,9 +2285,12 @@ def _update_cluster_status(cluster_name: str) -> Optional[Dict[str, Any]]:
                             -1,
                             autostop_lib.DEFAULT_AUTOSTOP_WAIT_FOR,
                             stream_logs=False)
-                    except exceptions.CommandError as e:
+                    except (exceptions.CommandError,
+                            grpc.FutureTimeoutError) as e:
                         success = False
-                        if e.returncode == 255:
+                        if isinstance(e, grpc.FutureTimeoutError) or (
+                                isinstance(e, exceptions.CommandError) and
+                                e.returncode == 255):
                             word = 'autostopped' if noun == 'autostop' else 'autodowned'
                             logger.debug(f'The cluster is likely {word}.')
                             reset_local_autostop = False
@@ -2329,10 +2340,14 @@ def _update_cluster_status(cluster_name: str) -> Optional[Dict[str, Any]]:
             # are only stored for an hour by default), so it is possible that
             # the previous event has a status reason, but now it does not.
             init_reason_regex = f'^Cluster is abnormal because {init_reason} .*'
+        log_message = f'Cluster is abnormal because {init_reason}'
+        if status_reason:
+            log_message += f' ({status_reason})'
+        log_message += '. Transitioned to INIT.'
         global_user_state.add_cluster_event(
             cluster_name,
             status_lib.ClusterStatus.INIT,
-            f'Cluster is abnormal because {init_reason} ({status_reason}). Transitioned to INIT.',
+            log_message,
             global_user_state.ClusterEventType.STATUS_CHANGE,
             nop_if_duplicate=True,
             duplicate_regex=init_reason_regex)
@@ -2345,10 +2360,10 @@ def _update_cluster_status(cluster_name: str) -> Optional[Dict[str, Any]]:
     # Now is_abnormal is False: either node_statuses is empty or all nodes are
     # STOPPED.
     backend = backends.CloudVmRayBackend()
-    backend.post_teardown_cleanup(handle, terminate=to_terminate, purge=False)
     global_user_state.add_cluster_event(
-        cluster_name, None, 'All nodes stopped, terminating cluster.',
+        cluster_name, None, 'All nodes terminated, cleaning up the cluster.',
         global_user_state.ClusterEventType.STATUS_CHANGE)
+    backend.post_teardown_cleanup(handle, terminate=to_terminate, purge=False)
     return global_user_state.get_cluster_from_name(cluster_name)
 
 

sky/backends/cloud_vm_ray_backend.py

@@ -1368,8 +1368,11 @@ class RetryingVmProvisioner(object):
         if not dryrun:
             os.makedirs(os.path.expanduser(self.log_dir), exist_ok=True)
             os.system(f'touch {log_path}')
+
         rich_utils.force_update_status(
-            ux_utils.spinner_message('Launching', log_path))
+            ux_utils.spinner_message('Launching',
+                                     log_path,
+                                     cluster_name=cluster_name))
 
         # Get previous cluster status
         cluster_exists = prev_cluster_status is not None
@@ -1539,6 +1542,7 @@ class RetryingVmProvisioner(object):
                 requested_resources=requested_resources,
                 ready=False,
                 is_managed=self._is_managed,
+                provision_log_path=log_abs_path,
             )
 
             # Add cluster event for actual provisioning start.
@@ -1684,7 +1688,9 @@ class RetryingVmProvisioner(object):
                 config_dict['handle'] = handle
                 logger.info(
                     ux_utils.finishing_message(
-                        f'Cluster launched: {cluster_name!r}.', log_path))
+                        f'Cluster launched: {cluster_name!r}.',
+                        log_path,
+                        cluster_name=cluster_name))
                 return config_dict
 
             # The cluster is not ready. We must perform error recording and/or
@@ -1818,7 +1824,8 @@ class RetryingVmProvisioner(object):
                 log_abs_path,
                 stream_logs=False,
                 start_streaming_at='Shared connection to',
-                line_processor=log_utils.RayUpLineProcessor(log_abs_path),
+                line_processor=log_utils.RayUpLineProcessor(
+                    log_abs_path, cluster_name=cluster_handle.cluster_name),
                 # Reduce BOTO_MAX_RETRIES from 12 to 5 to avoid long hanging
                 # time during 'ray up' if insufficient capacity occurs.
                 env=dict(
@@ -3120,7 +3127,9 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                         is_managed=self._is_managed)
                     log_path = os.path.join(self.log_dir, 'provision.log')
                     rich_utils.force_update_status(
-                        ux_utils.spinner_message('Launching', log_path))
+                        ux_utils.spinner_message('Launching',
+                                                 log_path,
+                                                 cluster_name=cluster_name))
                     config_dict = retry_provisioner.provision_with_retries(
                         task, to_provision_config, dryrun, stream_logs,
                         skip_unnecessary_provisioning)
@@ -3159,8 +3168,15 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                     # Do not remove the stopped cluster from the global state
                     # if failed to start.
                     if not e.no_failover:
+                        global_user_state.add_cluster_event(
+                            cluster_name,
+                            None,
+                            'Provision failed: ' + str(e),
+                            global_user_state.ClusterEventType.STATUS_CHANGE,
+                            nop_if_duplicate=True)
                         global_user_state.remove_cluster(cluster_name,
-                                                         terminate=True)
+                                                         terminate=True,
+                                                         remove_events=False)
                         usage_lib.messages.usage.update_final_cluster_status(
                             None)
                     logger.error(
@@ -3962,7 +3978,8 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
     def _teardown(self,
                   handle: CloudVmRayResourceHandle,
                   terminate: bool,
-                  purge: bool = False):
+                  purge: bool = False,
+                  explicitly_requested: bool = False):
         """Tear down or stop the cluster.
 
         Args:
@@ -4037,7 +4054,8 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                         # ClusterOwnerIdentityMismatchError. The argument/flag
                         # `purge` should bypass such ID mismatch errors.
                         refresh_cluster_status=(
-                            not is_identity_mismatch_and_purge))
+                            not is_identity_mismatch_and_purge),
+                        explicitly_requested=explicitly_requested)
                 if terminate:
                     lock.force_unlock()
                 break
@@ -4418,7 +4436,8 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                          purge: bool = False,
                          post_teardown_cleanup: bool = True,
                          refresh_cluster_status: bool = True,
-                         remove_from_db: bool = True) -> None:
+                         remove_from_db: bool = True,
+                         explicitly_requested: bool = False) -> None:
         """Teardown the cluster without acquiring the cluster status lock.
 
         NOTE: This method should not be called without holding the cluster
@@ -4482,7 +4501,8 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                            f'provision yaml so it '
                            'has not been provisioned. Skipped.')
             global_user_state.remove_cluster(handle.cluster_name,
-                                             terminate=terminate)
+                                             terminate=terminate,
+                                             remove_events=False)
             return
         log_path = os.path.join(os.path.expanduser(self.log_dir),
                                 'teardown.log')
@@ -4539,8 +4559,12 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                     raise
 
             if post_teardown_cleanup:
-                self.post_teardown_cleanup(handle, terminate, purge,
-                                           remove_from_db)
+                self.post_teardown_cleanup(
+                    handle,
+                    terminate,
+                    purge,
+                    remove_from_db,
+                    explicitly_requested=explicitly_requested)
             return
 
         if (isinstance(cloud, clouds.IBM) and terminate and
@@ -4640,7 +4664,8 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
                               terminate: bool,
                               purge: bool = False,
                               remove_from_db: bool = True,
-                              failover: bool = False) -> None:
+                              failover: bool = False,
+                              explicitly_requested: bool = False) -> None:
         """Cleanup local configs/caches and delete TPUs after teardown.
 
         This method will handle the following cleanup steps:
@@ -4819,7 +4844,8 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
 
         if not terminate or remove_from_db:
             global_user_state.remove_cluster(handle.cluster_name,
-                                             terminate=terminate)
+                                             terminate=terminate,
+                                             remove_events=explicitly_requested)
 
     def remove_cluster_config(self, handle: CloudVmRayResourceHandle) -> None:
         """Remove the YAML config of a cluster."""
@@ -4928,11 +4954,17 @@ class CloudVmRayBackend(backends.Backend['CloudVmRayResourceHandle']):
             # We cannot check if the cluster is autostopping.
             return False
         if handle.is_grpc_enabled:
-            request = autostopv1_pb2.IsAutostoppingRequest()
-            response = backend_utils.invoke_skylet_with_retries(
-                handle, lambda: SkyletClient(handle.get_grpc_channel()).
-                is_autostopping(request))
-            return response.is_autostopping
+            try:
+                request = autostopv1_pb2.IsAutostoppingRequest()
+                response = backend_utils.invoke_skylet_with_retries(
+                    handle, lambda: SkyletClient(handle.get_grpc_channel()).
+                    is_autostopping(request))
+                return response.is_autostopping
+            except Exception as e:  # pylint: disable=broad-except
+                # The cluster may have been terminated, causing the gRPC call
+                # to timeout and fail.
+                logger.debug(f'Failed to check if cluster is autostopping: {e}')
+                return False
         else:
             logger.info(
                 'Using legacy remote execution for is_autostopping on '

sky/backends/local_docker_backend.py

@@ -256,7 +256,9 @@ class LocalDockerBackend(backends.Backend['LocalDockerResourceHandle']):
                 logger.error(
                     'Unable to run container - nvidia runtime for docker not '
                     'found. Have you installed nvidia-docker on your machine?')
-            global_user_state.remove_cluster(cluster_name, terminate=True)
+            global_user_state.remove_cluster(cluster_name,
+                                             terminate=True,
+                                             remove_events=False)
             raise e
         self.containers[handle] = container
         logger.info(
@@ -323,7 +325,8 @@ class LocalDockerBackend(backends.Backend['LocalDockerResourceHandle']):
     def _teardown(self,
                   handle: LocalDockerResourceHandle,
                   terminate: bool,
-                  purge: bool = False):
+                  purge: bool = False,
+                  explicitly_requested: bool = False):
         """Teardown kills the container."""
         del purge  # Unused.
         if not terminate:
@@ -339,7 +342,9 @@ class LocalDockerBackend(backends.Backend['LocalDockerResourceHandle']):
             container.remove(force=True)
         cluster_name = handle.get_cluster_name()
 
-        global_user_state.remove_cluster(cluster_name, terminate=True)
+        global_user_state.remove_cluster(cluster_name,
+                                         terminate=True,
+                                         remove_events=explicitly_requested)
 
     # --- Utilities ---
 

sky/client/cli/command.py

@@ -60,6 +60,7 @@ from sky.client.cli import git
 from sky.data import storage_utils
 from sky.provision.kubernetes import constants as kubernetes_constants
 from sky.provision.kubernetes import utils as kubernetes_utils
+from sky.schemas.api import responses
 from sky.server import common as server_common
 from sky.server import constants as server_constants
 from sky.server.requests import requests
@@ -123,7 +124,7 @@ def _get_cluster_records_and_set_ssh_config(
     clusters: Optional[List[str]],
     refresh: common.StatusRefreshMode = common.StatusRefreshMode.NONE,
     all_users: bool = False,
-) -> List[dict]:
+) -> List[responses.StatusResponse]:
     """Returns a list of clusters that match the glob pattern.
 
     Args:
@@ -1562,7 +1563,7 @@ def _status_kubernetes(show_all: bool):
 
 
 def _show_endpoint(query_clusters: Optional[List[str]],
-                   cluster_records: List[Dict[str, Any]], ip: bool,
+                   cluster_records: List[responses.StatusResponse], ip: bool,
                    endpoints: bool, endpoint: Optional[int]) -> None:
     show_endpoints = endpoints or endpoint is not None
     show_single_endpoint = endpoint is not None
@@ -2171,6 +2172,10 @@ def queue(clusters: List[str], skip_finished: bool, all_users: bool):
 
 @cli.command()
 @flags.config_option(expose_value=False)
+@click.option('--provision',
+              is_flag=True,
+              default=False,
+              help='Stream the cluster provisioning logs (provision.log).')
 @click.option(
     '--sync-down',
     '-s',
@@ -2207,6 +2212,7 @@ def queue(clusters: List[str], skip_finished: bool, all_users: bool):
 def logs(
     cluster: str,
     job_ids: Tuple[str, ...],
+    provision: bool,
     sync_down: bool,
     status: bool,  # pylint: disable=redefined-outer-name
     follow: bool,
@@ -2236,6 +2242,11 @@ def logs(
     4. If the job fails or fetching the logs fails, the command will exit with
     a non-zero return code.
     """
+    if provision and (sync_down or status or job_ids):
+        raise click.UsageError(
+            '--provision cannot be combined with job log options '
+            '(--sync-down/--status/job IDs).')
+
     if sync_down and status:
         raise click.UsageError(
             'Both --sync_down and --status are specified '
@@ -2248,6 +2259,10 @@ def logs(
 
     job_ids = None if not job_ids else job_ids
 
+    if provision:
+        # Stream provision logs
+        sys.exit(sdk.tail_provision_logs(cluster, follow=follow, tail=tail))
+
     if sync_down:
         with rich_utils.client_status(
                 ux_utils.spinner_message('Downloading logs')):
@@ -4786,7 +4801,7 @@ def pool():
                 type=str,
                 nargs=-1,
                 **_get_shell_complete_args(_complete_file_name))
-@click.option('--pool-name',
+@click.option('--pool',
               '-p',
               default=None,
               type=str,
@@ -4808,7 +4823,7 @@ def pool():
 @usage_lib.entrypoint
 def jobs_pool_apply(
     pool_yaml: Tuple[str, ...],
-    pool_name: Optional[str],
+    pool: Optional[str],  # pylint: disable=redefined-outer-name
     workdir: Optional[str],
     infra: Optional[str],
     cloud: Optional[str],
@@ -4841,11 +4856,11 @@ def jobs_pool_apply(
     """
     cloud, region, zone = _handle_infra_cloud_region_zone_options(
         infra, cloud, region, zone)
-    if pool_name is None:
-        pool_name = serve_lib.generate_service_name(pool=True)
+    if pool is None:
+        pool = serve_lib.generate_service_name(pool=True)
 
     task = _generate_task_with_service(
-        service_name=pool_name,
+        service_name=pool,
         service_yaml_args=pool_yaml,
         workdir=workdir,
         cloud=cloud,
@@ -4882,7 +4897,7 @@ def jobs_pool_apply(
         dag.add(task)
 
     request_id = managed_jobs.pool_apply(task,
-                                         pool_name,
+                                         pool,
                                          mode=serve_lib.UpdateMode(mode),
                                          _need_confirmation=not yes)
     _async_call_or_wait(request_id, async_call, 'sky.jobs.pool_apply')
@@ -5120,7 +5135,7 @@ def _handle_serve_logs(
 @usage_lib.entrypoint
 # TODO(tian): Add default argument for this CLI if none of the flags are
 # specified.
-def pool_logs(
+def jobs_pool_logs(
     pool_name: str,
     follow: bool,
     controller: bool,
@@ -6037,7 +6052,7 @@ def api_logs(request_id: Optional[str], server_logs: bool,
     # server accepts log_path-only streaming.
     req_id = (server_common.RequestId[None](request_id)
               if request_id is not None else None)
-    sdk.stream_and_get(req_id, log_path, tail, follow=follow)
+    sdk.stream_and_get(req_id, log_path, tail, follow)
 
 
 @api.command('cancel', cls=_DocumentedCodeCommand)

sky/client/sdk.py

@@ -855,6 +855,56 @@ def tail_logs(cluster_name: str,
                            resumable=(tail == 0))
 
 
+@usage_lib.entrypoint
+@server_common.check_server_healthy_or_start
+@versions.minimal_api_version(17)
+@annotations.client_api
+@rest.retry_transient_errors()
+def tail_provision_logs(cluster_name: str,
+                        follow: bool = True,
+                        tail: int = 0,
+                        output_stream: Optional['io.TextIOBase'] = None) -> int:
+    """Tails the provisioning logs (provision.log) for a cluster.
+
+    Args:
+        cluster_name: name of the cluster.
+        follow: follow the logs.
+        tail: lines from end to tail.
+        output_stream: optional stream to write logs.
+    Returns:
+        Exit code 0 on streaming success; raises on HTTP error.
+    """
+    body = payloads.ClusterNameBody(cluster_name=cluster_name)
+    params = {
+        'follow': str(follow).lower(),
+        'tail': tail,
+    }
+    response = server_common.make_authenticated_request(
+        'POST',
+        '/provision_logs',
+        json=json.loads(body.model_dump_json()),
+        params=params,
+        stream=True,
+        timeout=(client_common.API_SERVER_REQUEST_CONNECTION_TIMEOUT_SECONDS,
+                 None))
+    # Log request is idempotent when tail is 0, thus can resume previous
+    # streaming point on retry.
+    # request_id=None here because /provision_logs does not create an async
+    # request. Instead, it streams a plain file from the server. This does NOT
+    # violate the stream_response doc warning about None in multi-user
+    # environments: we are not asking stream_response to select “the latest
+    # request”. We already have the HTTP response to stream; request_id=None
+    # merely disables the follow-up GET. It is also necessary for --no-follow
+    # to return cleanly after printing the tailed lines. If we provided a
+    # non-None request_id here, the get(request_id) in stream_response(
+    # would fail since /provision_logs does not create a request record.
+    stream_response(request_id=None,
+                    response=response,
+                    output_stream=output_stream,
+                    resumable=(tail == 0))
+    return 0
+
+
 @usage_lib.entrypoint
 @server_common.check_server_healthy_or_start
 @annotations.client_api
@@ -1322,7 +1372,7 @@ def status(
     cluster_names: Optional[List[str]] = None,
     refresh: common.StatusRefreshMode = common.StatusRefreshMode.NONE,
     all_users: bool = False,
-) -> server_common.RequestId[List[Dict[str, Any]]]:
+) -> server_common.RequestId[List[responses.StatusResponse]]:
     """Gets cluster statuses.
 
     If cluster_names is given, return those clusters. Otherwise, return all

sky/clouds/kubernetes.py

@@ -3,7 +3,6 @@ import os
 import re
 import subprocess
 import tempfile
-import typing
 from typing import Dict, Iterator, List, Optional, Set, Tuple, Union
 
 import colorama
@@ -11,6 +10,7 @@ import colorama
 from sky import catalog
 from sky import clouds
 from sky import exceptions
+from sky import resources as resources_lib
 from sky import sky_logging
 from sky import skypilot_config
 from sky.adaptors import kubernetes
@@ -31,10 +31,6 @@ from sky.utils import resources_utils
 from sky.utils import schemas
 from sky.utils import volume as volume_lib
 
-if typing.TYPE_CHECKING:
-    # Renaming to avoid shadowing variables.
-    from sky import resources as resources_lib
-
 logger = sky_logging.init_logger(__name__)
 
 # Namespace for SkyPilot resources shared across multiple tenants on the
@@ -773,7 +769,7 @@ class Kubernetes(clouds.Cloud):
 
     @staticmethod
     def _warn_on_disk_size(resources: 'resources_lib.Resources'):
-        if resources.disk_size is not None:
+        if resources.disk_size != resources_lib.DEFAULT_DISK_SIZE_GB:
             logger.info(f'{colorama.Style.DIM}Disk size {resources.disk_size} '
                         'is not supported by Kubernetes. '
                         'To add additional disk, use volumes.'

sky/clouds/nebius.py

@@ -442,7 +442,9 @@ class Nebius(clouds.Cloud):
         del workspace_config  # Unused
         sdk = nebius.sdk()
         profile_client = nebius.iam().ProfileServiceClient(sdk)
-        profile = profile_client.get(nebius.iam().GetProfileRequest()).wait()
+        profile = nebius.sync_call(
+            profile_client.get(nebius.iam().GetProfileRequest(),
+                               timeout=nebius.READ_TIMEOUT))
         if profile.user_profile is not None:
             if profile.user_profile.attributes is None:
                 raise exceptions.CloudUserIdentityError(

sky/core.py

@@ -25,6 +25,7 @@ from sky.clouds import cloud as sky_cloud
 from sky.jobs.server import core as managed_jobs_core
 from sky.provision.kubernetes import constants as kubernetes_constants
 from sky.provision.kubernetes import utils as kubernetes_utils
+from sky.schemas.api import responses
 from sky.skylet import autostop_lib
 from sky.skylet import constants
 from sky.skylet import job_lib
@@ -95,7 +96,7 @@ def status(
     cluster_names: Optional[Union[str, List[str]]] = None,
     refresh: common.StatusRefreshMode = common.StatusRefreshMode.NONE,
     all_users: bool = False,
-) -> List[Dict[str, Any]]:
+) -> List[responses.StatusResponse]:
     # NOTE(dev): Keep the docstring consistent between the Python API and CLI.
     """Gets cluster statuses.
 
@@ -171,7 +172,9 @@ def status(
     clusters = backend_utils.get_clusters(refresh=refresh,
                                           cluster_names=cluster_names,
                                           all_users=all_users)
-    return clusters
+    return [
+        responses.StatusResponse.model_validate(cluster) for cluster in clusters
+    ]
 
 
 def status_kubernetes(
@@ -593,7 +596,10 @@ def down(cluster_name: str, purge: bool = False) -> None:
 
     usage_lib.record_cluster_name_for_current_operation(cluster_name)
     backend = backend_utils.get_backend_from_handle(handle)
-    backend.teardown(handle, terminate=True, purge=purge)
+    backend.teardown(handle,
+                     terminate=True,
+                     purge=purge,
+                     explicitly_requested=True)
 
 
 @usage_lib.entrypoint

sky/dashboard/out/404.html

@@ -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-00c0a51d21157453.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-c2ea34fda4f1f8c8.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/_error-c66a4e8afc46f17b.js" defer=""></script><script src="/dashboard/_next/static/Y0eNlwi85qGRecLTin11y/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/Y0eNlwi85qGRecLTin11y/_ssgManifest.js" defer=""></script></head><body><div id="__next"></div><script id="__NEXT_DATA__" type="application/json">{"props":{"pageProps":{"statusCode":404}},"page":"/_error","query":{},"buildId":"Y0eNlwi85qGRecLTin11y","assetPrefix":"/dashboard","nextExport":true,"isFallback":false,"gip":true,"scriptLoader":[]}</script></body></html>
+<!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-b6987eb47888da9c.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/_error-c66a4e8afc46f17b.js" defer=""></script><script src="/dashboard/_next/static/yW7-Bc1l0EwIosbauU8LZ/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/yW7-Bc1l0EwIosbauU8LZ/_ssgManifest.js" defer=""></script></head><body><div id="__next"></div><script id="__NEXT_DATA__" type="application/json">{"props":{"pageProps":{"statusCode":404}},"page":"/_error","query":{},"buildId":"yW7-Bc1l0EwIosbauU8LZ","assetPrefix":"/dashboard","nextExport":true,"isFallback":false,"gip":true,"scriptLoader":[]}</script></body></html>