skypilot-nightly 1.0.0.dev20250812__py3-none-any.whl → 1.0.0.dev20250815__py3-none-any.whl

sky/client/cli/command.py

@@ -275,65 +275,6 @@ def _merge_env_vars(env_dict: Optional[Dict[str, str]],
     return list(env_dict.items())
 
 
-def _format_job_ids_str(job_ids: List[int], max_length: int = 30) -> str:
-    """Format job IDs string with ellipsis if too long.
-
-    Args:
-        job_ids: List of job IDs to format.
-        max_length: Maximum length of the output string.
-
-    Returns:
-        Formatted string like "11,12,...,2017,2018" if truncated,
-        or the full string if it fits within max_length.
-    """
-    if not job_ids:
-        return ''
-
-    # Convert all to strings
-    job_strs = [str(job_id) for job_id in job_ids]
-    full_str = ','.join(job_strs)
-
-    # If it fits, return as is
-    if len(full_str) <= max_length:
-        return full_str
-
-    if len(job_strs) <= 2:
-        return full_str  # Can't truncate further
-
-    # Need to truncate with ellipsis
-    ellipsis = '...'
-
-    # Start with minimum: first and last
-    start_count = 1
-    end_count = 1
-
-    while start_count + end_count < len(job_strs):
-        # Try adding one more to start
-        if start_count + 1 + end_count < len(job_strs):
-            start_part = ','.join(job_strs[:start_count + 1])
-            end_part = ','.join(job_strs[-end_count:])
-            candidate = f'{start_part},{ellipsis},{end_part}'
-            if len(candidate) <= max_length:
-                start_count += 1
-                continue
-
-        # Try adding one more to end
-        if start_count + end_count + 1 < len(job_strs):
-            start_part = ','.join(job_strs[:start_count])
-            end_part = ','.join(job_strs[-(end_count + 1):])
-            candidate = f'{start_part},{ellipsis},{end_part}'
-            if len(candidate) <= max_length:
-                end_count += 1
-                continue
-
-        # Can't add more
-        break
-
-    start_part = ','.join(job_strs[:start_count])
-    end_part = ','.join(job_strs[-end_count:])
-    return f'{start_part},{ellipsis},{end_part}'
-
-
 def _complete_cluster_name(ctx: click.Context, param: click.Parameter,
                            incomplete: str) -> List[str]:
     """Handle shell completion for cluster names."""
@@ -1187,11 +1128,15 @@ def launch(
             raise ValueError(f'{backend_name} backend is not supported.')
 
     if task.service is not None:
+        noun = 'pool' if task.service.pool else 'service'
+        capnoun = noun.capitalize()
+        sysname = 'Jobs Worker Pool' if task.service.pool else 'SkyServe'
+        cmd = 'sky jobs pool apply' if task.service.pool else 'sky serve up'
         logger.info(
-            f'{colorama.Fore.YELLOW}Service section will be ignored when using '
-            f'`sky launch`. {colorama.Style.RESET_ALL}\n{colorama.Fore.YELLOW}'
-            'To spin up a service, use SkyServe CLI: '
-            f'{colorama.Style.RESET_ALL}{colorama.Style.BRIGHT}sky serve up'
+            f'{colorama.Fore.YELLOW}{capnoun} section will be ignored when '
+            f'using `sky launch`. {colorama.Style.RESET_ALL}\n'
+            f'{colorama.Fore.YELLOW}To spin up a {noun}, use {sysname} CLI: '
+            f'{colorama.Style.RESET_ALL}{colorama.Style.BRIGHT}{cmd}'
             f'{colorama.Style.RESET_ALL}')
 
     request_id = sdk.launch(
@@ -2226,6 +2171,10 @@ 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',
@@ -2262,6 +2211,7 @@ 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,
@@ -2291,6 +2241,11 @@ 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 '
@@ -2303,6 +2258,10 @@ 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')):
@@ -2981,15 +2940,15 @@ def _hint_or_raise_for_down_jobs_controller(controller_name: str,
     controller = controller_utils.Controllers.from_name(controller_name)
     assert controller is not None, controller_name
 
-    # TODO(tian): We also need to check pools after we allow running pools on
-    # jobs controller.
     with rich_utils.client_status(
-            '[bold cyan]Checking for in-progress managed jobs[/]'):
+            '[bold cyan]Checking for in-progress managed jobs and pools[/]'):
         try:
             request_id = managed_jobs.queue(refresh=False,
                                             skip_finished=True,
                                             all_users=True)
             managed_jobs_ = sdk.stream_and_get(request_id)
+            request_id_pools = managed_jobs.pool_status(pool_names=None)
+            pools_ = sdk.stream_and_get(request_id_pools)
         except exceptions.ClusterNotUpError as e:
             if controller.value.connection_error_hint in str(e):
                 with ux_utils.print_exception_no_traceback():
@@ -3004,6 +2963,7 @@ def _hint_or_raise_for_down_jobs_controller(controller_name: str,
             # the controller being STOPPED or being firstly launched, i.e.,
             # there is no in-prgress managed jobs.
             managed_jobs_ = []
+            pools_ = []
         except exceptions.InconsistentConsolidationModeError:
             # If this error is raised, it means the user switched to the
             # consolidation mode but the previous controller cluster is still
@@ -3021,6 +2981,8 @@ def _hint_or_raise_for_down_jobs_controller(controller_name: str,
                                                 skip_finished=True,
                                                 all_users=True)
                 managed_jobs_ = sdk.stream_and_get(request_id)
+                request_id_pools = managed_jobs.pool_status(pool_names=None)
+                pools_ = sdk.stream_and_get(request_id_pools)
 
     msg = (f'{colorama.Fore.YELLOW}WARNING: Tearing down the managed '
            'jobs controller. Please be aware of the following:'
@@ -3042,9 +3004,23 @@ def _hint_or_raise_for_down_jobs_controller(controller_name: str,
         else:
             with ux_utils.print_exception_no_traceback():
                 raise exceptions.NotSupportedError(msg)
+    elif pools_:
+        pool_names = ', '.join([pool['name'] for pool in pools_])
+        if purge:
+            logger.warning('--purge is set, ignoring the in-progress pools. '
+                           'This could cause leaked clusters!')
+        else:
+            msg = (f'{colorama.Fore.YELLOW}WARNING: Tearing down the managed '
+                   'jobs controller is not supported, as it is currently '
+                   f'hosting the following pools: {pool_names}. Please '
+                   'terminate the pools first with '
+                   f'{colorama.Style.BRIGHT}sky jobs pool down -a'
+                   f'{colorama.Style.RESET_ALL}.')
+            with ux_utils.print_exception_no_traceback():
+                raise exceptions.NotSupportedError(msg)
     else:
-        click.echo(' * No in-progress managed jobs found. It should be safe to '
-                   'terminate (see caveats above).')
+        click.echo(' * No in-progress managed jobs or running pools found. It '
+                   'should be safe to terminate (see caveats above).')
 
 
 def _hint_or_raise_for_down_sky_serve_controller(controller_name: str,
@@ -4509,8 +4485,8 @@ def jobs_launch(
         if print_setup_fm_warning:
             click.secho(
                 f'{colorama.Fore.YELLOW}setup/file_mounts/storage_mounts'
-                ' will be ignored in pool. To update a pool, please '
-                f'use `sky pool apply {pool} pool.yaml`. '
+                ' will be ignored when submit jobs to pool. To update a pool, '
+                f'please use `sky jobs pool apply {pool} new-pool.yaml`. '
                 f'{colorama.Style.RESET_ALL}')
 
     # Optimize info is only show if _need_confirmation.
@@ -4537,7 +4513,9 @@ def jobs_launch(
                                                 controller=False)
             sys.exit(returncode)
         else:
-            job_ids_str = _format_job_ids_str(job_ids)
+            # TODO(tian): This can be very long. Considering have a "group id"
+            # and query all job ids with the same group id.
+            job_ids_str = ','.join(map(str, job_ids))
             click.secho(
                 f'Jobs submitted with IDs: {colorama.Fore.CYAN}'
                 f'{job_ids_str}{colorama.Style.RESET_ALL}.'
@@ -4822,7 +4800,7 @@ 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,
@@ -4844,7 +4822,7 @@ 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],
@@ -4877,11 +4855,11 @@ 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,
@@ -4918,7 +4896,7 @@ 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')
@@ -5156,7 +5134,7 @@ 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,
@@ -6073,7 +6051,7 @@ 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)
@@ -6212,16 +6190,15 @@ def api_info():
     """Shows the SkyPilot API server URL."""
     url = server_common.get_server_url()
     api_server_info = sdk.api_info()
-    api_server_user = api_server_info.get('user')
+    api_server_user = api_server_info.user
     if api_server_user is not None:
-        user = models.User(id=api_server_user['id'],
-                           name=api_server_user['name'])
+        user = api_server_user
     else:
         user = models.User.get_current_user()
     click.echo(f'Using SkyPilot API server and dashboard: {url}\n'
-               f'{ux_utils.INDENT_SYMBOL}Status: {api_server_info["status"]}, '
-               f'commit: {api_server_info["commit"]}, '
-               f'version: {api_server_info["version"]}\n'
+               f'{ux_utils.INDENT_SYMBOL}Status: {api_server_info.status}, '
+               f'commit: {api_server_info.commit}, '
+               f'version: {api_server_info.version}\n'
                f'{ux_utils.INDENT_LAST_SYMBOL}User: {user.name} ({user.id})')
 
 

sky/client/common.py

@@ -32,7 +32,7 @@ if typing.TYPE_CHECKING:
     import requests
 
     import sky
-    import sky.dag as dag_lib
+    from sky import dag as dag_lib
 else:
     httpx = adaptors_common.LazyImport('httpx')
     requests = adaptors_common.LazyImport('requests')

sky/client/sdk.py

@@ -30,6 +30,7 @@ from sky import skypilot_config
 from sky.adaptors import common as adaptors_common
 from sky.client import common as client_common
 from sky.client import oauth as oauth_lib
+from sky.schemas.api import responses
 from sky.server import common as server_common
 from sky.server import rest
 from sky.server import versions
@@ -66,8 +67,8 @@ if typing.TYPE_CHECKING:
 
     import sky
     from sky import backends
+    from sky import catalog
     from sky import models
-    import sky.catalog
     from sky.provision.kubernetes import utils as kubernetes_utils
     from sky.skylet import job_lib
 else:
@@ -234,7 +235,7 @@ def list_accelerators(
     require_price: bool = True,
     case_sensitive: bool = True
 ) -> server_common.RequestId[Dict[str,
-                                  List['sky.catalog.common.InstanceTypeInfo']]]:
+                                  List['catalog.common.InstanceTypeInfo']]]:
     """Lists the names of all accelerators offered by Sky.
 
     This will include all accelerators offered by Sky, including those
@@ -479,11 +480,11 @@ def launch(
             This option works in conjunction with ``idle_minutes_to_autostop``.
             Choices:
 
-            1. "jobs_and_ssh" (default) - Wait for all jobs to complete
-               AND all SSH sessions to disconnect.
-            2. "jobs" - Wait for all jobs to complete.
-            3. "none" - Stop immediately after idle time expires,
-               regardless of running jobs or SSH connections.
+            1. "jobs_and_ssh" (default) - Wait for in-progress jobs and SSH
+               connections to finish.
+            2. "jobs" - Only wait for in-progress jobs.
+            3. "none" - Wait for nothing; autostop right after
+               ``idle_minutes_to_autostop``.
         dryrun: if True, do not actually launch the cluster.
         down: Tear down the cluster after all jobs finish (successfully or
             abnormally). If --idle-minutes-to-autostop is also set, the
@@ -854,6 +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
@@ -935,11 +986,11 @@ def start(
             This option works in conjunction with ``idle_minutes_to_autostop``.
             Choices:
 
-            1. "jobs_and_ssh" (default) - Wait for all jobs to complete
-               AND all SSH sessions to disconnect.
-            2. "jobs" - Wait for all jobs to complete.
-            3. "none" - Stop immediately after idle time expires,
-               regardless of running jobs or SSH connections.
+            1. "jobs_and_ssh" (default) - Wait for in-progress jobs and SSH
+               connections to finish.
+            2. "jobs" - Only wait for in-progress jobs.
+            3. "none" - Wait for nothing; autostop right after
+               ``idle_minutes_to_autostop``.
         retry_until_up: whether to retry launching the cluster until it is
             up.
         down: Autodown the cluster: tear down the cluster after specified
@@ -1118,11 +1169,10 @@ def autostop(
             This option works in conjunction with ``idle_minutes``.
             Choices:
 
-            1. "jobs_and_ssh" (default) - Wait for all jobs to complete
-               AND all SSH sessions to disconnect.
-            2. "jobs" - Wait for all jobs to complete.
-            3. "none" - Stop immediately after idle time expires,
-               regardless of running jobs or SSH connections.
+            1. "jobs_and_ssh" (default) - Wait for in-progress jobs and SSH
+               connections to finish.
+            2. "jobs" - Only wait for in-progress jobs.
+            3. "none" - Wait for nothing; autostop right after ``idle_minutes``.
         down: if true, use autodown (tear down the cluster; non-restartable),
             rather than autostop (restartable).
 
@@ -2059,7 +2109,7 @@ def api_status(
 @usage_lib.entrypoint
 @server_common.check_server_healthy_or_start
 @annotations.client_api
-def api_info() -> Dict[str, Any]:
+def api_info() -> responses.APIHealthResponse:
     """Gets the server's status, commit and version.
 
     Returns:
@@ -2084,7 +2134,7 @@ def api_info() -> Dict[str, Any]:
     """
     response = server_common.make_authenticated_request('GET', '/api/health')
     response.raise_for_status()
-    return response.json()
+    return responses.APIHealthResponse(**response.json())
 
 
 @usage_lib.entrypoint

sky/client/sdk_async.py

@@ -20,13 +20,14 @@ import colorama
 
 from sky import admin_policy
 from sky import backends
+from sky import catalog
 from sky import exceptions
 from sky import models
 from sky import sky_logging
-import sky.catalog
 from sky.client import common as client_common
 from sky.client import sdk
 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 rest
 from sky.server.requests import payloads
@@ -301,7 +302,7 @@ async def list_accelerators(
     require_price: bool = True,
     case_sensitive: bool = True,
     stream_logs: Optional[StreamConfig] = DEFAULT_STREAM_CONFIG
-) -> Dict[str, List[sky.catalog.common.InstanceTypeInfo]]:
+) -> Dict[str, List[catalog.common.InstanceTypeInfo]]:
     """Async version of list_accelerators() that lists the names of all
     accelerators offered by Sky."""
     request_id = await context_utils.to_thread(sdk.list_accelerators, gpus_only,
@@ -345,7 +346,7 @@ async def optimize(
         admin_policy_request_options: Optional[
             admin_policy.RequestOptions] = None,
         stream_logs: Optional[StreamConfig] = DEFAULT_STREAM_CONFIG
-) -> sky.dag.Dag:
+) -> 'sky.Dag':
     """Async version of optimize() that finds the best execution plan for the
       given DAG."""
     request_id = await context_utils.to_thread(sdk.optimize, dag, minimize,
@@ -786,7 +787,7 @@ async def dashboard(starting_page: Optional[str] = None) -> None:
 
 @usage_lib.entrypoint
 @annotations.client_api
-async def api_info() -> Dict[str, Any]:
+async def api_info() -> responses.APIHealthResponse:
     """Async version of api_info() that gets the server's status, commit and
       version."""
     return await context_utils.to_thread(sdk.api_info)

sky/clouds/aws.py

@@ -65,6 +65,8 @@ _CREDENTIAL_FILES = [
 ]
 
 DEFAULT_AMI_GB = 45
+DEFAULT_SSH_USER = 'ubuntu'
+DEFAULT_ROOT_DEVICE_NAME = '/dev/sda1'
 
 # Temporary measure, as deleting per-cluster SGs is too slow.
 # See https://github.com/skypilot-org/skypilot/pull/742.
@@ -343,6 +345,44 @@ class AWS(clouds.Cloud):
                 raise ValueError(image_not_found_message) from None
         return image_size
 
+    @classmethod
+    @annotations.lru_cache(scope='request', maxsize=1)
+    def get_image_root_device_name(cls, image_id: str,
+                                   region: Optional[str]) -> str:
+        if image_id.startswith('skypilot:'):
+            return DEFAULT_ROOT_DEVICE_NAME
+        assert region is not None, (image_id, region)
+        image_not_found_message = (
+            f'Image {image_id!r} not found in AWS region {region}.\n'
+            f'To find AWS AMI IDs: https://docs.aws.amazon.com/cli/latest/reference/ec2/describe-images.html#examples\n'  # pylint: disable=line-too-long
+            'Example: ami-0729d913a335efca7')
+        try:
+            client = aws.client('ec2', region_name=region)
+            image_info = client.describe_images(ImageIds=[image_id]).get(
+                'Images', [])
+            if not image_info:
+                with ux_utils.print_exception_no_traceback():
+                    raise ValueError(image_not_found_message)
+            image = image_info[0]
+            if 'RootDeviceName' not in image:
+                logger.warning(f'Image {image_id!r} does not have a root '
+                               f'device name. '
+                               f'Using {DEFAULT_ROOT_DEVICE_NAME}.')
+                return DEFAULT_ROOT_DEVICE_NAME
+            return image['RootDeviceName']
+        except (aws.botocore_exceptions().NoCredentialsError,
+                aws.botocore_exceptions().ProfileNotFound):
+            # Fallback to default root device name if no credentials are
+            # available.
+            # The credentials issue will be caught when actually provisioning
+            # the instance and appropriate errors will be raised there.
+            logger.warning(f'No credentials available for region {region}. '
+                           f'Using {DEFAULT_ROOT_DEVICE_NAME}.')
+            return DEFAULT_ROOT_DEVICE_NAME
+        except aws.botocore_exceptions().ClientError:
+            with ux_utils.print_exception_no_traceback():
+                raise ValueError(image_not_found_message) from None
+
     @classmethod
     def get_zone_shell_cmd(cls) -> Optional[str]:
         # The command for getting the current zone is from:
@@ -466,6 +506,15 @@ class AWS(clouds.Cloud):
         image_id = self._get_image_id(image_id_to_use, region_name,
                                       resources.instance_type)
 
+        root_device_name = self.get_image_root_device_name(
+            image_id, region_name)
+
+        ssh_user = skypilot_config.get_effective_region_config(
+            cloud='aws',
+            region=region_name,
+            keys=('ssh_user',),
+            default_value=DEFAULT_SSH_USER)
+
         disk_encrypted = skypilot_config.get_effective_region_config(
             cloud='aws',
             region=region_name,
@@ -509,6 +558,8 @@ class AWS(clouds.Cloud):
             'region': region_name,
             'zones': ','.join(zone_names),
             'image_id': image_id,
+            'root_device_name': root_device_name,
+            'ssh_user': ssh_user,
             'security_group': security_group,
             'security_group_managed_by_skypilot':
                 str(security_group != user_security_group).lower(),
@@ -1080,7 +1131,7 @@ class AWS(clouds.Cloud):
 
         image_name = f'skypilot-{cluster_name.display_name}-{int(time.time())}'
 
-        status = provision_lib.query_instances('AWS',
+        status = provision_lib.query_instances('AWS', cluster_name.display_name,
                                                cluster_name.name_on_cloud,
                                                {'region': region})
         instance_ids = list(status.keys())

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
@@ -771,11 +767,25 @@ class Kubernetes(clouds.Cloud):
 
         return deploy_vars
 
+    @staticmethod
+    def _warn_on_disk_size(resources: 'resources_lib.Resources'):
+        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.'
+                        f'{colorama.Style.RESET_ALL}')
+        if resources.disk_tier is not None:
+            logger.info(f'{colorama.Style.DIM}Disk tier {resources.disk_tier} '
+                        'is not supported by Kubernetes. '
+                        'To add additional disk, use volumes.'
+                        f'{colorama.Style.RESET_ALL}')
+
     def _get_feasible_launchable_resources(
         self, resources: 'resources_lib.Resources'
     ) -> 'resources_utils.FeasibleResources':
         # TODO(zhwu): This needs to be updated to return the correct region
         # (context) that has enough resources.
+        self._warn_on_disk_size(resources)
         fuzzy_candidate_list: List[str] = []
         if resources.instance_type is not None:
             assert resources.is_launchable(), resources

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/dag.py

@@ -24,6 +24,7 @@ class Dag:
         self.graph = nx.DiGraph()
         self.name: Optional[str] = None
         self.policy_applied: bool = False
+        self.pool: Optional[str] = None
 
     def add(self, task: 'task.Task') -> None:
         self.graph.add_node(task)

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-7fd0cf9dbecff10f.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-491a4d699d95e808.js" defer=""></script><script src="/dashboard/_next/static/chunks/pages/_error-c66a4e8afc46f17b.js" defer=""></script><script src="/dashboard/_next/static/Fuy7OzApYTUMz2QgoP7dP/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/Fuy7OzApYTUMz2QgoP7dP/_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":"Fuy7OzApYTUMz2QgoP7dP","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-aba778a6d6eb496d.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/I-djf3wB8zZl_bI67BOyZ/_buildManifest.js" defer=""></script><script src="/dashboard/_next/static/I-djf3wB8zZl_bI67BOyZ/_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":"I-djf3wB8zZl_bI67BOyZ","assetPrefix":"/dashboard","nextExport":true,"isFallback":false,"gip":true,"scriptLoader":[]}</script></body></html>