skypilot-nightly 1.0.0.dev20250212__py3-none-any.whl → 1.0.0.dev20250214__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 = '1fe3fab0e7a3242f32039d55b456603350dc4196'
+_SKYPILOT_COMMIT_SHA = '7170a91ca2605e15290689b675bc7640999f4aa1'
 
 
 def _get_git_commit():
@@ -35,7 +35,7 @@ def _get_git_commit():
 
 
 __commit__ = _get_git_commit()
-__version__ = '1.0.0.dev20250212'
+__version__ = '1.0.0.dev20250214'
 __root_dir__ = os.path.dirname(os.path.abspath(__file__))
 
 

sky/clouds/kubernetes.py

@@ -464,7 +464,9 @@ class Kubernetes(clouds.Cloud):
         # CPU resources on the node instead within the pod.
         custom_ray_options = {
             'object-store-memory': 500000000,
-            'num-cpus': str(int(cpus)),
+            # 'num-cpus' must be an integer, but we should not set it to 0 if
+            # cpus is <1.
+            'num-cpus': str(max(int(cpus), 1)),
         }
         deploy_vars = {
             'instance_type': resources.instance_type,

sky/clouds/runpod.py

@@ -177,6 +177,11 @@ class RunPod(clouds.Cloud):
         hourly_cost = self.instance_type_to_hourly_cost(
             instance_type=instance_type, use_spot=use_spot)
 
+        # default to root
+        docker_username_for_runpod = (resources.docker_username_for_runpod
+                                      if resources.docker_username_for_runpod
+                                      is not None else 'root')
+
         return {
             'instance_type': instance_type,
             'custom_resources': custom_resources,
@@ -184,6 +189,7 @@ class RunPod(clouds.Cloud):
             'image_id': image_id,
             'use_spot': use_spot,
             'bid_per_gpu': str(hourly_cost),
+            'docker_username_for_runpod': docker_username_for_runpod,
         }
 
     def _get_feasible_launchable_resources(

sky/clouds/service_catalog/data_fetchers/fetch_lambda_cloud.py

@@ -28,6 +28,7 @@ REGIONS = [
     'asia-northeast-2',
     'us-east-1',
     'us-east-2',
+    'us-east-3',
     'us-west-2',
     'us-west-1',
     'us-south-1',

sky/data/storage.py

@@ -354,7 +354,8 @@ class AbstractStore:
                                              metadata.is_sky_managed),
             sync_on_reconstruction=override_args.get('sync_on_reconstruction',
                                                      True),
-            # backward compatibility
+            # Backward compatibility
+            # TODO: remove the hasattr check after v0.11.0
             _bucket_sub_path=override_args.get(
                 '_bucket_sub_path',
                 metadata._bucket_sub_path  # pylint: disable=protected-access
@@ -1462,6 +1463,8 @@ class S3Store(AbstractStore):
                 set to True, the directory is created in the bucket root and
                 contents are uploaded to it.
         """
+        sub_path = (f'/{self._bucket_sub_path}'
+                    if self._bucket_sub_path else '')
 
         def get_file_sync_command(base_dir_path, file_names):
             includes = ' '.join([
@@ -1469,8 +1472,6 @@ class S3Store(AbstractStore):
                 for file_name in file_names
             ])
             base_dir_path = shlex.quote(base_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = ('aws s3 sync --no-follow-symlinks --exclude="*" '
                             f'{includes} {base_dir_path} '
                             f's3://{self.name}{sub_path}')
@@ -1485,8 +1486,6 @@ class S3Store(AbstractStore):
                 for file_name in excluded_list
             ])
             src_dir_path = shlex.quote(src_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = (f'aws s3 sync --no-follow-symlinks {excludes} '
                             f'{src_dir_path} '
                             f's3://{self.name}{sub_path}/{dest_dir_name}')
@@ -1500,7 +1499,7 @@ class S3Store(AbstractStore):
 
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
-        sync_path = f'{source_message} -> s3://{self.name}/'
+        sync_path = f'{source_message} -> s3://{self.name}{sub_path}/'
         with rich_utils.safe_status(
                 ux_utils.spinner_message(f'Syncing {sync_path}',
                                          log_path=log_path)):
@@ -1959,11 +1958,13 @@ class GcsStore(AbstractStore):
         copy_list = '\n'.join(
             os.path.abspath(os.path.expanduser(p)) for p in source_path_list)
         gsutil_alias, alias_gen = data_utils.get_gsutil_command()
+        sub_path = (f'/{self._bucket_sub_path}'
+                    if self._bucket_sub_path else '')
         sync_command = (f'{alias_gen}; echo "{copy_list}" | {gsutil_alias} '
-                        f'cp -e -n -r -I gs://{self.name}')
+                        f'cp -e -n -r -I gs://{self.name}{sub_path}')
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
-        sync_path = f'{source_message} -> gs://{self.name}/'
+        sync_path = f'{source_message} -> gs://{self.name}{sub_path}/'
         with rich_utils.safe_status(
                 ux_utils.spinner_message(f'Syncing {sync_path}',
                                          log_path=log_path)):
@@ -1995,13 +1996,13 @@ class GcsStore(AbstractStore):
                 set to True, the directory is created in the bucket root and
                 contents are uploaded to it.
         """
+        sub_path = (f'/{self._bucket_sub_path}'
+                    if self._bucket_sub_path else '')
 
         def get_file_sync_command(base_dir_path, file_names):
             sync_format = '|'.join(file_names)
             gsutil_alias, alias_gen = data_utils.get_gsutil_command()
             base_dir_path = shlex.quote(base_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = (f'{alias_gen}; {gsutil_alias} '
                             f'rsync -e -x \'^(?!{sync_format}$).*\' '
                             f'{base_dir_path} gs://{self.name}{sub_path}')
@@ -2014,8 +2015,6 @@ class GcsStore(AbstractStore):
             excludes = '|'.join(excluded_list)
             gsutil_alias, alias_gen = data_utils.get_gsutil_command()
             src_dir_path = shlex.quote(src_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = (f'{alias_gen}; {gsutil_alias} '
                             f'rsync -e -r -x \'({excludes})\' {src_dir_path} '
                             f'gs://{self.name}{sub_path}/{dest_dir_name}')
@@ -2029,7 +2028,7 @@ class GcsStore(AbstractStore):
 
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
-        sync_path = f'{source_message} -> gs://{self.name}/'
+        sync_path = f'{source_message} -> gs://{self.name}{sub_path}/'
         with rich_utils.safe_status(
                 ux_utils.spinner_message(f'Syncing {sync_path}',
                                          log_path=log_path)):
@@ -2307,15 +2306,24 @@ class AzureBlobStore(AbstractStore):
             An instance of AzureBlobStore.
         """
         assert isinstance(metadata, AzureBlobStore.AzureBlobStoreMetadata)
-        return cls(name=override_args.get('name', metadata.name),
-                   storage_account_name=override_args.get(
-                       'storage_account', metadata.storage_account_name),
-                   source=override_args.get('source', metadata.source),
-                   region=override_args.get('region', metadata.region),
-                   is_sky_managed=override_args.get('is_sky_managed',
-                                                    metadata.is_sky_managed),
-                   sync_on_reconstruction=override_args.get(
-                       'sync_on_reconstruction', True))
+        # TODO: this needs to be kept in sync with the abstract
+        # AbstractStore.from_metadata.
+        return cls(
+            name=override_args.get('name', metadata.name),
+            storage_account_name=override_args.get(
+                'storage_account', metadata.storage_account_name),
+            source=override_args.get('source', metadata.source),
+            region=override_args.get('region', metadata.region),
+            is_sky_managed=override_args.get('is_sky_managed',
+                                             metadata.is_sky_managed),
+            sync_on_reconstruction=override_args.get('sync_on_reconstruction',
+                                                     True),
+            # Backward compatibility
+            # TODO: remove the hasattr check after v0.11.0
+            _bucket_sub_path=override_args.get(
+                '_bucket_sub_path',
+                metadata._bucket_sub_path  # pylint: disable=protected-access
+            ) if hasattr(metadata, '_bucket_sub_path') else None)
 
     def get_metadata(self) -> AzureBlobStoreMetadata:
         return self.AzureBlobStoreMetadata(
@@ -2795,6 +2803,8 @@ class AzureBlobStore(AbstractStore):
                 set to True, the directory is created in the bucket root and
                 contents are uploaded to it.
         """
+        container_path = (f'{self.container_name}/{self._bucket_sub_path}'
+                          if self._bucket_sub_path else self.container_name)
 
         def get_file_sync_command(base_dir_path, file_names) -> str:
             # shlex.quote is not used for file_names as 'az storage blob sync'
@@ -2803,8 +2813,6 @@ class AzureBlobStore(AbstractStore):
             includes_list = ';'.join(file_names)
             includes = f'--include-pattern "{includes_list}"'
             base_dir_path = shlex.quote(base_dir_path)
-            container_path = (f'{self.container_name}/{self._bucket_sub_path}'
-                              if self._bucket_sub_path else self.container_name)
             sync_command = (f'az storage blob sync '
                             f'--account-name {self.storage_account_name} '
                             f'--account-key {self.storage_account_key} '
@@ -2822,18 +2830,17 @@ class AzureBlobStore(AbstractStore):
                 [file_name.rstrip('*') for file_name in excluded_list])
             excludes = f'--exclude-path "{excludes_list}"'
             src_dir_path = shlex.quote(src_dir_path)
-            container_path = (f'{self.container_name}/{self._bucket_sub_path}'
-                              if self._bucket_sub_path else
-                              f'{self.container_name}')
             if dest_dir_name:
-                container_path = f'{container_path}/{dest_dir_name}'
+                dest_dir_name = f'/{dest_dir_name}'
+            else:
+                dest_dir_name = ''
             sync_command = (f'az storage blob sync '
                             f'--account-name {self.storage_account_name} '
                             f'--account-key {self.storage_account_key} '
                             f'{excludes} '
                             '--delete-destination false '
                             f'--source {src_dir_path} '
-                            f'--container {container_path}')
+                            f'--container {container_path}{dest_dir_name}')
             return sync_command
 
         # Generate message for upload
@@ -2844,7 +2851,7 @@ class AzureBlobStore(AbstractStore):
             source_message = source_path_list[0]
         container_endpoint = data_utils.AZURE_CONTAINER_URL.format(
             storage_account_name=self.storage_account_name,
-            container_name=self.name)
+            container_name=container_path)
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
         sync_path = f'{source_message} -> {container_endpoint}/'
@@ -3238,6 +3245,8 @@ class R2Store(AbstractStore):
                 set to True, the directory is created in the bucket root and
                 contents are uploaded to it.
         """
+        sub_path = (f'/{self._bucket_sub_path}'
+                    if self._bucket_sub_path else '')
 
         def get_file_sync_command(base_dir_path, file_names):
             includes = ' '.join([
@@ -3246,8 +3255,6 @@ class R2Store(AbstractStore):
             ])
             endpoint_url = cloudflare.create_endpoint()
             base_dir_path = shlex.quote(base_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = ('AWS_SHARED_CREDENTIALS_FILE='
                             f'{cloudflare.R2_CREDENTIALS_PATH} '
                             'aws s3 sync --no-follow-symlinks --exclude="*" '
@@ -3267,8 +3274,6 @@ class R2Store(AbstractStore):
             ])
             endpoint_url = cloudflare.create_endpoint()
             src_dir_path = shlex.quote(src_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = ('AWS_SHARED_CREDENTIALS_FILE='
                             f'{cloudflare.R2_CREDENTIALS_PATH} '
                             f'aws s3 sync --no-follow-symlinks {excludes} '
@@ -3286,7 +3291,7 @@ class R2Store(AbstractStore):
 
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
-        sync_path = f'{source_message} -> r2://{self.name}/'
+        sync_path = f'{source_message} -> r2://{self.name}{sub_path}/'
         with rich_utils.safe_status(
                 ux_utils.spinner_message(f'Syncing {sync_path}',
                                          log_path=log_path)):
@@ -3710,6 +3715,8 @@ class IBMCosStore(AbstractStore):
                 set to True, the directory is created in the bucket root and
                 contents are uploaded to it.
         """
+        sub_path = (f'/{self._bucket_sub_path}'
+                    if self._bucket_sub_path else '')
 
         def get_dir_sync_command(src_dir_path, dest_dir_name) -> str:
             """returns an rclone command that copies a complete folder
@@ -3731,8 +3738,6 @@ class IBMCosStore(AbstractStore):
             # .git directory is excluded from the sync
             # wrapping src_dir_path with "" to support path with spaces
             src_dir_path = shlex.quote(src_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = (
                 'rclone copy --exclude ".git/*" '
                 f'{src_dir_path} '
@@ -3763,8 +3768,6 @@ class IBMCosStore(AbstractStore):
                 for file_name in file_names
             ])
             base_dir_path = shlex.quote(base_dir_path)
-            sub_path = (f'/{self._bucket_sub_path}'
-                        if self._bucket_sub_path else '')
             sync_command = (
                 'rclone copy '
                 f'{includes} {base_dir_path} '
@@ -3779,7 +3782,8 @@ class IBMCosStore(AbstractStore):
 
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
-        sync_path = f'{source_message} -> cos://{self.region}/{self.name}/'
+        sync_path = (
+            f'{source_message} -> cos://{self.region}/{self.name}{sub_path}/')
         with rich_utils.safe_status(
                 ux_utils.spinner_message(f'Syncing {sync_path}',
                                          log_path=log_path)):
@@ -4178,15 +4182,21 @@ class OciStore(AbstractStore):
                 set to True, the directory is created in the bucket root and
                 contents are uploaded to it.
         """
+        sub_path = (f'{self._bucket_sub_path}/'
+                    if self._bucket_sub_path else '')
 
         @oci.with_oci_env
         def get_file_sync_command(base_dir_path, file_names):
             includes = ' '.join(
                 [f'--include "{file_name}"' for file_name in file_names])
+            prefix_arg = ''
+            if sub_path:
+                prefix_arg = f'--object-prefix "{sub_path.strip("/")}"'
             sync_command = (
                 'oci os object bulk-upload --no-follow-symlinks --overwrite '
                 f'--bucket-name {self.name} --namespace-name {self.namespace} '
                 f'--region {self.region} --src-dir "{base_dir_path}" '
+                f'{prefix_arg} '
                 f'{includes}')
 
             return sync_command
@@ -4207,7 +4217,8 @@ class OciStore(AbstractStore):
             sync_command = (
                 'oci os object bulk-upload --no-follow-symlinks --overwrite '
                 f'--bucket-name {self.name} --namespace-name {self.namespace} '
-                f'--region {self.region} --object-prefix "{dest_dir_name}" '
+                f'--region {self.region} '
+                f'--object-prefix "{sub_path}{dest_dir_name}" '
                 f'--src-dir "{src_dir_path}" {excludes}')
 
             return sync_command
@@ -4220,7 +4231,7 @@ class OciStore(AbstractStore):
 
         log_path = sky_logging.generate_tmp_logging_file_path(
             _STORAGE_LOG_FILE_NAME)
-        sync_path = f'{source_message} -> oci://{self.name}/'
+        sync_path = f'{source_message} -> oci://{self.name}/{sub_path}'
         with rich_utils.safe_status(
                 ux_utils.spinner_message(f'Syncing {sync_path}',
                                          log_path=log_path)):

sky/provision/provisioner.py

@@ -450,6 +450,13 @@ def _post_provision_setup(
         logger.info(f'{indent_str}{colorama.Style.DIM}{vm_str}{plural} {verb} '
                     f'up.{colorama.Style.RESET_ALL}')
 
+        # It's promised by the cluster config that docker_config does not
+        # exist for docker-native clouds, i.e. they provide docker containers
+        # instead of full VMs, like Kubernetes and RunPod, as it requires some
+        # special handlings to run docker inside their docker virtualization.
+        # For their Docker image settings, we do them when provisioning the
+        # cluster. See provision/{cloud}/instance.py:get_cluster_info for more
+        # details.
         if docker_config:
             status.update(
                 ux_utils.spinner_message(

sky/provision/runpod/utils.py

@@ -186,7 +186,7 @@ def delete_pod_template(template_name: str) -> None:
         runpod.runpod.api.graphql.run_graphql_query(
             f'mutation {{deleteTemplate(templateName: "{template_name}")}}')
     except runpod.runpod.error.QueryError as e:
-        logger.warning(f'Failed to delete template {template_name}: {e}'
+        logger.warning(f'Failed to delete template {template_name}: {e} '
                        'Please delete it manually.')
 
 
@@ -195,8 +195,9 @@ def delete_register_auth(registry_auth_id: str) -> None:
     try:
         runpod.runpod.delete_container_registry_auth(registry_auth_id)
     except runpod.runpod.error.QueryError as e:
-        logger.warning(f'Failed to delete registry auth {registry_auth_id}: {e}'
-                       'Please delete it manually.')
+        logger.warning(
+            f'Failed to delete registry auth {registry_auth_id}: {e} '
+            'Please delete it manually.')
 
 
 def _create_template_for_docker_login(

sky/resources.py

@@ -67,6 +67,7 @@ class Resources:
         # Internal use only.
         # pylint: disable=invalid-name
         _docker_login_config: Optional[docker_utils.DockerLoginConfig] = None,
+        _docker_username_for_runpod: Optional[str] = None,
         _is_image_managed: Optional[bool] = None,
         _requires_fuse: Optional[bool] = None,
         _cluster_config_overrides: Optional[Dict[str, Any]] = None,
@@ -148,6 +149,9 @@ class Resources:
           _docker_login_config: the docker configuration to use. This includes
             the docker username, password, and registry server. If None, skip
             docker login.
+          _docker_username_for_runpod: the login username for the docker
+            containers. This is used by RunPod to set the ssh user for the
+            docker containers.
           _requires_fuse: whether the task requires FUSE mounting support. This
             is used internally by certain cloud implementations to do additional
             setup for FUSE mounting. This flag also safeguards against using
@@ -234,6 +238,12 @@ class Resources:
 
         self._docker_login_config = _docker_login_config
 
+        # TODO(andyl): This ctor param seems to be unused.
+        # We always use `Task.set_resources` and `Resources.copy` to set the
+        # `docker_username_for_runpod`. But to keep the consistency with
+        # `_docker_login_config`, we keep it here.
+        self._docker_username_for_runpod = _docker_username_for_runpod
+
         self._requires_fuse = _requires_fuse
 
         self._cluster_config_overrides = _cluster_config_overrides
@@ -479,6 +489,10 @@ class Resources:
     def requires_fuse(self, value: Optional[bool]) -> None:
         self._requires_fuse = value
 
+    @property
+    def docker_username_for_runpod(self) -> Optional[str]:
+        return self._docker_username_for_runpod
+
     def _set_cpus(
         self,
         cpus: Union[None, int, float, str],
@@ -1065,6 +1079,10 @@ class Resources:
         cloud_specific_variables = self.cloud.make_deploy_resources_variables(
             self, cluster_name, region, zones, num_nodes, dryrun)
 
+        # TODO(andyl): Should we print some warnings if users' envs share
+        # same names with the cloud specific variables, but not enabled
+        # since it's not on the particular cloud?
+
         # Docker run options
         docker_run_options = skypilot_config.get_nested(
             ('docker', 'run_options'),
@@ -1277,6 +1295,9 @@ class Resources:
             labels=override.pop('labels', self.labels),
             _docker_login_config=override.pop('_docker_login_config',
                                               self._docker_login_config),
+            _docker_username_for_runpod=override.pop(
+                '_docker_username_for_runpod',
+                self._docker_username_for_runpod),
             _is_image_managed=override.pop('_is_image_managed',
                                            self._is_image_managed),
             _requires_fuse=override.pop('_requires_fuse', self._requires_fuse),
@@ -1438,6 +1459,8 @@ class Resources:
         resources_fields['labels'] = config.pop('labels', None)
         resources_fields['_docker_login_config'] = config.pop(
             '_docker_login_config', None)
+        resources_fields['_docker_username_for_runpod'] = config.pop(
+            '_docker_username_for_runpod', None)
         resources_fields['_is_image_managed'] = config.pop(
             '_is_image_managed', None)
         resources_fields['_requires_fuse'] = config.pop('_requires_fuse', None)
@@ -1486,6 +1509,9 @@ class Resources:
         if self._docker_login_config is not None:
             config['_docker_login_config'] = dataclasses.asdict(
                 self._docker_login_config)
+        if self._docker_username_for_runpod is not None:
+            config['_docker_username_for_runpod'] = (
+                self._docker_username_for_runpod)
         add_if_not_none('_cluster_config_overrides',
                         self._cluster_config_overrides)
         if self._is_image_managed is not None:

sky/skylet/constants.py

@@ -110,6 +110,8 @@ DOCKER_LOGIN_ENV_VARS = {
     DOCKER_SERVER_ENV_VAR,
 }
 
+RUNPOD_DOCKER_USERNAME_ENV_VAR = 'SKYPILOT_RUNPOD_DOCKER_USERNAME'
+
 # Commands for disable GPU ECC, which can improve the performance of the GPU
 # for some workloads by 30%. This will only be applied when a user specify
 # `nvidia_gpus.disable_ecc: true` in ~/.sky/config.yaml.

sky/task.py

@@ -121,6 +121,9 @@ def _check_docker_login_config(task_envs: Dict[str, str]) -> bool:
 
     If any of the docker login env vars is set, all of them must be set.
 
+    Returns:
+        True if there is a valid docker login config in task_envs.
+        False otherwise.
     Raises:
         ValueError: if any of the docker login env vars is set, but not all of
             them are set.
@@ -168,6 +171,23 @@ def _with_docker_login_config(
     return type(resources)(new_resources)
 
 
+def _with_docker_username_for_runpod(
+    resources: Union[Set['resources_lib.Resources'],
+                     List['resources_lib.Resources']],
+    task_envs: Dict[str, str],
+) -> Union[Set['resources_lib.Resources'], List['resources_lib.Resources']]:
+    docker_username_for_runpod = task_envs.get(
+        constants.RUNPOD_DOCKER_USERNAME_ENV_VAR)
+
+    # We should not call r.copy() if docker_username_for_runpod is None,
+    # to prevent `DummyResources` instance becoming a `Resources` instance.
+    if docker_username_for_runpod is None:
+        return resources
+    return (type(resources)(
+        r.copy(_docker_username_for_runpod=docker_username_for_runpod)
+        for r in resources))
+
+
 class Task:
     """Task: a computation to be run on the cloud."""
 
@@ -582,6 +602,8 @@ class Task:
         if _check_docker_login_config(self._envs):
             self.resources = _with_docker_login_config(self.resources,
                                                        self._envs)
+        self.resources = _with_docker_username_for_runpod(
+            self.resources, self._envs)
         return self
 
     @property
@@ -647,6 +669,9 @@ class Task:
             resources = {resources}
         # TODO(woosuk): Check if the resources are None.
         self.resources = _with_docker_login_config(resources, self.envs)
+        # Only have effect on RunPod.
+        self.resources = _with_docker_username_for_runpod(
+            self.resources, self.envs)
 
         # Evaluate if the task requires FUSE and set the requires_fuse flag
         for _, storage_obj in self.storage_mounts.items():

sky/templates/runpod-ray.yml.j2

@@ -25,7 +25,7 @@ provider:
   {%- endif %}
 
 auth:
-  ssh_user: root
+  ssh_user: {{docker_username_for_runpod}}
   ssh_private_key: {{ssh_private_key}}
 
 available_node_types:

{skypilot_nightly-1.0.0.dev20250212.dist-info → skypilot_nightly-1.0.0.dev20250214.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: skypilot-nightly
-Version: 1.0.0.dev20250212
+Version: 1.0.0.dev20250214
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0
@@ -184,7 +184,9 @@ Dynamic: summary
 
 ----
 :fire: *News* :fire:
-- [Jan 2025] Launch and Serve **[DeepSeek-R1](https://github.com/deepseek-ai/DeepSeek-R1)** and **[Janus](https://github.com/deepseek-ai/DeepSeek-Janus)** on Kubernetes or Any Cloud: [**R1 example**](./llm/deepseek-r1/) and [**Janus example**](./llm/deepseek-janus/)
+- [Feb 2025] Run and Serve DeepSeek-R1 671B using SkyPilot and SGLang with high throughput: [**example**](./llm/deepseek-r1/)
+- [Jan 2025] Prepare and Serve Large-Scale Image Search with **Vector Database**: [**blog post**](https://blog.skypilot.co/large-scale-vector-database/) [**example**](./examples/vector_database/) 
+- [Jan 2025] Launch and Serve distilled models from **[DeepSeek-R1](https://github.com/deepseek-ai/DeepSeek-R1)** and **[Janus](https://github.com/deepseek-ai/DeepSeek-Janus)** on Kubernetes or Any Cloud: [**R1 example**](./llm/deepseek-r1-distilled/) and [**Janus example**](./llm/deepseek-janus/)
 - [Oct 2024] :tada: **SkyPilot crossed 1M+ downloads** :tada:: Thank you to our community! [**Twitter/X**](https://x.com/skypilot_org/status/1844770841718067638)
 - [Sep 2024] Point, Launch and Serve **Llama 3.2** on Kubernetes or Any Cloud: [**example**](./llm/llama-3_2/)
 - [Sep 2024] Run and deploy [**Pixtral**](./llm/pixtral), the first open-source multimodal model from Mistral AI.
@@ -342,7 +344,7 @@ Runnable examples:
   - [LocalGPT](./llm/localgpt)
   - [Falcon](./llm/falcon)
   - Add yours here & see more in [`llm/`](./llm)!
-- Framework examples: [PyTorch DDP](https://github.com/skypilot-org/skypilot/blob/master/examples/resnet_distributed_torch.yaml), [DeepSpeed](./examples/deepspeed-multinode/sky.yaml), [JAX/Flax on TPU](https://github.com/skypilot-org/skypilot/blob/master/examples/tpu/tpuvm_mnist.yaml), [Stable Diffusion](https://github.com/skypilot-org/skypilot/tree/master/examples/stable_diffusion), [Detectron2](https://github.com/skypilot-org/skypilot/blob/master/examples/detectron2_docker.yaml), [Distributed](https://github.com/skypilot-org/skypilot/blob/master/examples/resnet_distributed_tf_app.py) [TensorFlow](https://github.com/skypilot-org/skypilot/blob/master/examples/resnet_app_storage.yaml), [Ray Train](examples/distributed_ray_train/ray_train.yaml), [NeMo](https://github.com/skypilot-org/skypilot/blob/master/examples/nemo/), [programmatic grid search](https://github.com/skypilot-org/skypilot/blob/master/examples/huggingface_glue_imdb_grid_search_app.py), [Docker](https://github.com/skypilot-org/skypilot/blob/master/examples/docker/echo_app.yaml), [Cog](https://github.com/skypilot-org/skypilot/blob/master/examples/cog/), [Unsloth](https://github.com/skypilot-org/skypilot/blob/master/examples/unsloth/unsloth.yaml), [Ollama](https://github.com/skypilot-org/skypilot/blob/master/llm/ollama), [llm.c](https://github.com/skypilot-org/skypilot/tree/master/llm/gpt-2), [Airflow](./examples/airflow/training_workflow) and [many more (`examples/`)](./examples).
+- Framework examples: [Vector Database](./examples/vector_database/), [PyTorch DDP](https://github.com/skypilot-org/skypilot/blob/master/examples/resnet_distributed_torch.yaml), [DeepSpeed](./examples/deepspeed-multinode/sky.yaml), [JAX/Flax on TPU](https://github.com/skypilot-org/skypilot/blob/master/examples/tpu/tpuvm_mnist.yaml), [Stable Diffusion](https://github.com/skypilot-org/skypilot/tree/master/examples/stable_diffusion), [Detectron2](https://github.com/skypilot-org/skypilot/blob/master/examples/detectron2_docker.yaml), [Distributed](https://github.com/skypilot-org/skypilot/blob/master/examples/resnet_distributed_tf_app.py) [TensorFlow](https://github.com/skypilot-org/skypilot/blob/master/examples/resnet_app_storage.yaml), [Ray Train](examples/distributed_ray_train/ray_train.yaml), [NeMo](https://github.com/skypilot-org/skypilot/blob/master/examples/nemo/), [programmatic grid search](https://github.com/skypilot-org/skypilot/blob/master/examples/huggingface_glue_imdb_grid_search_app.py), [Docker](https://github.com/skypilot-org/skypilot/blob/master/examples/docker/echo_app.yaml), [Cog](https://github.com/skypilot-org/skypilot/blob/master/examples/cog/), [Unsloth](https://github.com/skypilot-org/skypilot/blob/master/examples/unsloth/unsloth.yaml), [Ollama](https://github.com/skypilot-org/skypilot/blob/master/llm/ollama), [llm.c](https://github.com/skypilot-org/skypilot/tree/master/llm/gpt-2), [Airflow](./examples/airflow/training_workflow) and [many more (`examples/`)](./examples).
 
 Case Studies and Integrations: [Community Spotlights](https://blog.skypilot.co/community/)
 

{skypilot_nightly-1.0.0.dev20250212.dist-info → skypilot_nightly-1.0.0.dev20250214.dist-info}/RECORD

@@ -1,4 +1,4 @@
-sky/__init__.py,sha256=GRvhpT8lUvIyIPLyC2cqv9RiI6hZ0_iMY56YNDpjPbs,5560
+sky/__init__.py,sha256=hnw0sI17qKlRS50e2Ux9neBg1F8HcuZMLdlV7S9ziIE,5560
 sky/admin_policy.py,sha256=hPo02f_A32gCqhUueF0QYy1fMSSKqRwYEg_9FxScN_s,3248
 sky/authentication.py,sha256=MNc9uHnvQ1EsEl8SsrYcYCGbxcnDbR6gaRCXVNd5RZE,22338
 sky/check.py,sha256=xzLlxUkBCrzpOho8lw65EvKLPl_b9lA2nteF5MSYbDQ,10885
@@ -10,11 +10,11 @@ sky/exceptions.py,sha256=SEhRubPlk-crkflPC5P_Z085iLrSd3UScYwc790QwYw,9378
 sky/execution.py,sha256=vNUE9Z8hCSQeil7h3kdote2r6nkbrGXSqqmK6ru594Q,28453
 sky/global_user_state.py,sha256=cTwltMCDIIBaapuGgARxFwpDJDCiKKyVW-PP_qtWuCA,30241
 sky/optimizer.py,sha256=H5cpKELOQmnFpox0QXMB4P7jGhJxzXog4Ht_TYJaGuA,59758
-sky/resources.py,sha256=D3jteQxKOUydoNm7VDl90p02dwP3RpbO3gqNcl4dpOI,70327
+sky/resources.py,sha256=W7VO5nTizr-KIhOamOs7oSwmBGLjQZhQM6DoYbiAOsg,71648
 sky/sky_logging.py,sha256=7Zk9mL1TDxFkGsy3INMBKYlqsbognVGSMzAsHZdZlhw,5891
 sky/skypilot_config.py,sha256=FN93hSG-heQCHBnemlIK2TwrJngKbpx4vMXNUzPIzV8,9087
 sky/status_lib.py,sha256=J7Jb4_Dz0v2T64ttOdyUgpokvl4S0sBJrMfH7Fvo51A,1457
-sky/task.py,sha256=zri5_Ghh5-fjDf2rl997ZmL4IlXetW9u9XXJIRUJ3Qg,51353
+sky/task.py,sha256=Z74bBkOx1bFmGMoQRD3qbYxIZ5qi2AC5htY1KIsmvT0,52394
 sky/adaptors/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sky/adaptors/aws.py,sha256=FNNC8B-iqSSiCLPiWGK_PLm1R-Kt4yI5JPIpdE0QJxQ,7565
 sky/adaptors/azure.py,sha256=yjM8nAPW-mlSXfmA8OmJNnSIrZ9lQx2-GxiI-TIVrwE,21910
@@ -51,11 +51,11 @@ sky/clouds/do.py,sha256=od4gMTrs2W5IkaDDr_oexOSdIOqn94vKq2U_QZcrpRk,11311
 sky/clouds/fluidstack.py,sha256=u2I6jXEtTqgqRWi2EafMsKqc8VkUq1cR6CSDUvk72_U,12407
 sky/clouds/gcp.py,sha256=6QOnefFsYiLCcnajjduLHsayqJ641bBu42jPTpvy7Mc,55007
 sky/clouds/ibm.py,sha256=0ArRTQx1_DpTNGByFhukzFedEDzmVjBsGiiques1bQ0,21447
-sky/clouds/kubernetes.py,sha256=oZg4Lpn2ZBikyc5NTJIziUPEY0xs2mtz546ButhkZ7g,31541
+sky/clouds/kubernetes.py,sha256=ocf8ZUlMbOcPb-n8JrSFix9rH75g089sC1JAd84JUXQ,31653
 sky/clouds/lambda_cloud.py,sha256=42AmcN2X_wdBMuAw606nR_pQCBAy5QFiAo711_WRqDE,12672
 sky/clouds/oci.py,sha256=VpPxpMJv52QePVdwdK9EuiMyqjp70dk8_rgUVv5Y-2w,27028
 sky/clouds/paperspace.py,sha256=F0Sj1RcqRb5fPjrr8qbdeY-JdfAHcRPc902pZOi4erw,10889
-sky/clouds/runpod.py,sha256=Wtaarp27_LTu5_E2agC7tTr2vhN1D4sblr2vZTT4vBI,11580
+sky/clouds/runpod.py,sha256=4gZTbUO4I8bzio5x9Km42ZIujZzCx2eszdkN0Mz5fqE,11893
 sky/clouds/scp.py,sha256=JHyMqkAAqr9lJq79IVjj3rU1g-ZCCGLZTJEzIhYsw7c,15845
 sky/clouds/vast.py,sha256=vQV489qkZMfDtt_SnXParPY49gkgKx5LZAEOsk65kIo,11231
 sky/clouds/vsphere.py,sha256=rrNf6_uHy4ukjHwaN35XVh2-Xj9k43-QGQkiEXyHYJk,12273
@@ -85,7 +85,7 @@ sky/clouds/service_catalog/data_fetchers/fetch_azure.py,sha256=7YVnoGDGGZI2TK02b
 sky/clouds/service_catalog/data_fetchers/fetch_cudo.py,sha256=52P48lvWN0s1ArjeLPeLemPRpxjSRcHincRle0nqdm4,3440
 sky/clouds/service_catalog/data_fetchers/fetch_fluidstack.py,sha256=yKuAFbjBRNz_e2RNNDT_aHHAuKQ86Ac7GKgIie5O6Pg,7273
 sky/clouds/service_catalog/data_fetchers/fetch_gcp.py,sha256=HLxdxA9DMSi19mgpVM_cERV4o-xh_tJ9vmkGm1wOaIE,30868
-sky/clouds/service_catalog/data_fetchers/fetch_lambda_cloud.py,sha256=MN54h0CAGPHQAeF2eTmuESq3b0-d1kDARRUM6OkivCk,4962
+sky/clouds/service_catalog/data_fetchers/fetch_lambda_cloud.py,sha256=Bi5ta91p4SkFCoaEJUPKPjDB0FZ24DMR4NcKFpKCkxU,4979
 sky/clouds/service_catalog/data_fetchers/fetch_vast.py,sha256=zR9icM3ty5C8tGw13pQbsBtQQMgG4kl1j_jSGqqrgOA,4741
 sky/clouds/service_catalog/data_fetchers/fetch_vsphere.py,sha256=Opp2r3KSzXPtwk3lKNbO8IX9QzjoRSwy1kW3jPjtS1c,21453
 sky/clouds/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -98,7 +98,7 @@ sky/data/__init__.py,sha256=Nhaf1NURisXpZuwWANa2IuCyppIuc720FRwqSE2oEwY,184
 sky/data/data_transfer.py,sha256=wixC4_3_JaeJFdGKOp-O5ulcsMugDSgrCR0SnPpugGc,8946
 sky/data/data_utils.py,sha256=HjcgMDuWRR_fNQ9gjuROi9GgPVvTGApiJwxGtdb2_UU,28860
 sky/data/mounting_utils.py,sha256=tJHBPEDP1Wg_r3oSGBwFhMDLnPCMPSFRz26O0QkDd0Y,14908
-sky/data/storage.py,sha256=CWVKnHhdzXw1biPbRqYizkyVexL_OCELuJCqtd4hit4,204094
+sky/data/storage.py,sha256=vlipyOOxw43W6t5O9ccL53OYWAxPVMjF16qlfbZhdQo,204431
 sky/data/storage_utils.py,sha256=cM3kxlffYE7PnJySDu8huyUsMX_JYsf9uer8r5OYsjo,9556
 sky/jobs/__init__.py,sha256=ObZcz3lL1ip8JcmR6gbfZ4RMMfXJJdsnuU2zLQUb8jY,1546
 sky/jobs/constants.py,sha256=9kIdpwWNI9zWKQO39LTg9spUMGl5Iqx4ByIjRlic7Hw,1893
@@ -118,7 +118,7 @@ sky/provision/docker_utils.py,sha256=ENm0LkyrYWic3Ikyacho8X5uDMvGsbkZQsb6kNH1DuI
 sky/provision/instance_setup.py,sha256=YBFOwZQLBzpUjYoVQcX0KItej1rCBRWM23Dw9lg_q24,24386
 sky/provision/logging.py,sha256=yZWgejrFBhhRjAtvFu5N5bRXIMK5TuwNjp1vKQqz2pw,2103
 sky/provision/metadata_utils.py,sha256=LrxeV4wD2QPzNdXV_npj8q-pr35FatxBBjF_jSbpOT0,4013
-sky/provision/provisioner.py,sha256=ZOgFOO0NB4QZVPwd4qikRqi615Bq67n0Vcl3cTDVxNE,29153
+sky/provision/provisioner.py,sha256=UuMO2wp98Nk1nue25TtIaMU79Aj14v9IoIAFO_CUy9w,29636
 sky/provision/aws/__init__.py,sha256=mxq8PeWJqUtalDozTNpbtENErRZ1ktEs8uf2aG9UUgU,731
 sky/provision/aws/config.py,sha256=-4mr5uxgsl_8eLm_4DfP8JurZGSysGuY0iDeBTHnX5Q,25943
 sky/provision/aws/instance.py,sha256=3-R8ohuN8ooNh2Fqqb7-c4vNFy1xsw2GQF4PHg3APhE,40843
@@ -172,7 +172,7 @@ sky/provision/paperspace/utils.py,sha256=uOmxbDKjV6skFizC4gYXSxRuEqso5ck2kF7MbtN
 sky/provision/runpod/__init__.py,sha256=6HYvHI27EaLrX1SS0vWVhdLu5HDBeZCdvAeDJuwM5pk,556
 sky/provision/runpod/config.py,sha256=9ulZJVL7nHuxhTdoj8D7lNn7SdicJ5zc6FIcHIG9tcg,321
 sky/provision/runpod/instance.py,sha256=FNalpTQMvnmACdtFsGvVPVhEkzdqrlmd_pExn33bIQ8,10358
-sky/provision/runpod/utils.py,sha256=2L4MUrFi4rQOP26IeAkPNQ9oWGPJhH5fvz7NAbL3nNQ,12421
+sky/provision/runpod/utils.py,sha256=F3zsxPV3IY5C73J-zNZEfcXL8bsZBgNWzlmtSO6qOqI,12425
 sky/provision/runpod/api/__init__.py,sha256=eJwjPeQZ5B7chf4-Bl4YeI2Uo9aLX4M1rr2NmPk89_E,112
 sky/provision/runpod/api/commands.py,sha256=oh77PS0H0wZudHV8II9ceRuaFQ8FN4NJ4S3-6_PeqPM,4238
 sky/provision/runpod/api/pods.py,sha256=GMwxgNr9NnHPfyh2Y9b8S_vLhrLY4h7LybFBBQNAyfw,4948
@@ -215,7 +215,7 @@ sky/skylet/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sky/skylet/attempt_skylet.py,sha256=GZ6ITjjA0m-da3IxXXfoHR6n4pjp3X3TOXUqVvSrV0k,2136
 sky/skylet/autostop_lib.py,sha256=JPDHmByuhoNYXSUHl-OnyeJUkOFWn7gDM1FrS7Kr3E8,4478
 sky/skylet/configs.py,sha256=UtnpmEL0F9hH6PSjhsps7xgjGZ6qzPOfW1p2yj9tSng,1887
-sky/skylet/constants.py,sha256=EUSW4yH59eqBDLMIdmQWIYd3nAJBFoUeo5v9MGiginI,16057
+sky/skylet/constants.py,sha256=bTvojyYHLhUxMI-xD8d6Fj6snQiK4-PLw-baRFKCcfQ,16125
 sky/skylet/events.py,sha256=__7bt6Z8q2W1vwTQv4yug-oAXDwSf8zBeRxb8HFM36U,12792
 sky/skylet/job_lib.py,sha256=Rk-C069cusJIRXsks8xqCb016JSt7GlpU7LrpX0qFJk,42785
 sky/skylet/log_lib.py,sha256=oFEBd85vDYFrIyyZKekH30yc4rRYILC0F0o-COQ64oE,20445
@@ -256,7 +256,7 @@ sky/templates/lambda-ray.yml.j2,sha256=HyvO_tX2vxwSsc4IFVSqGuIbjLMk0bevP9bcxb8ZQ
 sky/templates/local-ray.yml.j2,sha256=FNHeyHF6nW9nU9QLIZceUWfvrFTTcO51KqhTnYCEFaA,1185
 sky/templates/oci-ray.yml.j2,sha256=92dvXGaUd2Kwep9fgTjOsAPJiBLr8GQTjy7pFvuPAyE,4562
 sky/templates/paperspace-ray.yml.j2,sha256=HQjZNamrB_a4fOMCxQXSVdV5JIHtbGtAE0JzEO8uuVQ,4021
-sky/templates/runpod-ray.yml.j2,sha256=bUiF4Y_EkCA_GKLtTzPXbajdL-NOUiJ38Pe4dZd2dys,4284
+sky/templates/runpod-ray.yml.j2,sha256=95392Jvk-PwrEHIF2C5i-EnaqC31nTq4UTYZuQxUe1k,4310
 sky/templates/scp-ray.yml.j2,sha256=I9u8Ax-lit-d6UrCC9BVU8avst8w1cwK6TrzZBcz_JM,5608
 sky/templates/sky-serve-controller.yaml.j2,sha256=W4i1-OGRU2WDvauLC4EDXcYrNxj7mzRFSvSqzAKfehc,2020
 sky/templates/vast-ray.yml.j2,sha256=KaZLBJfI6FzAVRVq0NNM0_SN0RQUrDIehnJJ_LnvwnY,2990
@@ -298,9 +298,9 @@ sky/utils/kubernetes/k8s_gpu_labeler_job.yaml,sha256=k0TBoQ4zgf79-sVkixKSGYFHQ7Z
 sky/utils/kubernetes/k8s_gpu_labeler_setup.yaml,sha256=VLKT2KKimZu1GDg_4AIlIt488oMQvhRZWwsj9vBbPUg,3812
 sky/utils/kubernetes/rsync_helper.sh,sha256=h4YwrPFf9727CACnMJvF3EyK_0OeOYKKt4su_daKekw,1256
 sky/utils/kubernetes/ssh_jump_lifecycle_manager.py,sha256=Kq1MDygF2IxFmu9FXpCxqucXLmeUrvs6OtRij6XTQbo,6554
-skypilot_nightly-1.0.0.dev20250212.dist-info/LICENSE,sha256=emRJAvE7ngL6x0RhQvlns5wJzGI3NEQ_WMjNmd9TZc4,12170
-skypilot_nightly-1.0.0.dev20250212.dist-info/METADATA,sha256=rkJIHWHxQtacqsQPb5SZ7XHCGiXMvMBzXNPupXqi4sU,21397
-skypilot_nightly-1.0.0.dev20250212.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-skypilot_nightly-1.0.0.dev20250212.dist-info/entry_points.txt,sha256=StA6HYpuHj-Y61L2Ze-hK2IcLWgLZcML5gJu8cs6nU4,36
-skypilot_nightly-1.0.0.dev20250212.dist-info/top_level.txt,sha256=qA8QuiNNb6Y1OF-pCUtPEr6sLEwy2xJX06Bd_CrtrHY,4
-skypilot_nightly-1.0.0.dev20250212.dist-info/RECORD,,
+skypilot_nightly-1.0.0.dev20250214.dist-info/LICENSE,sha256=emRJAvE7ngL6x0RhQvlns5wJzGI3NEQ_WMjNmd9TZc4,12170
+skypilot_nightly-1.0.0.dev20250214.dist-info/METADATA,sha256=jxTYT5S0cFhCCQ6C77MoNQIW2yEJCxgX1Nz4LfPoUgI,21800
+skypilot_nightly-1.0.0.dev20250214.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+skypilot_nightly-1.0.0.dev20250214.dist-info/entry_points.txt,sha256=StA6HYpuHj-Y61L2Ze-hK2IcLWgLZcML5gJu8cs6nU4,36
+skypilot_nightly-1.0.0.dev20250214.dist-info/top_level.txt,sha256=qA8QuiNNb6Y1OF-pCUtPEr6sLEwy2xJX06Bd_CrtrHY,4
+skypilot_nightly-1.0.0.dev20250214.dist-info/RECORD,,