skypilot-nightly 1.0.0.dev20241228__py3-none-any.whl → 1.0.0.dev20241230__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 = '138679859b9844a8737f8dff1bf5a739e77e96c4'
+_SKYPILOT_COMMIT_SHA = '7e40bcdce7437f601bb07e6eaf9fd954efdd12c6'
 
 
 def _get_git_commit():
@@ -35,7 +35,7 @@ def _get_git_commit():
 
 
 __commit__ = _get_git_commit()
-__version__ = '1.0.0.dev20241228'
+__version__ = '1.0.0.dev20241230'
 __root_dir__ = os.path.dirname(os.path.abspath(__file__))
 
 

sky/adaptors/oci.py

@@ -1,9 +1,11 @@
 """Oracle OCI cloud adaptor"""
 
+import functools
 import logging
 import os
 
 from sky.adaptors import common
+from sky.clouds.utils import oci_utils
 
 # Suppress OCI circuit breaker logging before lazy import, because
 # oci modules prints additional message during imports, i.e., the
@@ -30,10 +32,16 @@ def get_config_file() -> str:
 
 def get_oci_config(region=None, profile='DEFAULT'):
     conf_file_path = get_config_file()
+    if not profile or profile == 'DEFAULT':
+        config_profile = oci_utils.oci_config.get_profile()
+    else:
+        config_profile = profile
+
     oci_config = oci.config.from_file(file_location=conf_file_path,
-                                      profile_name=profile)
+                                      profile_name=config_profile)
     if region is not None:
         oci_config['region'] = region
+
     return oci_config
 
 
@@ -54,6 +62,29 @@ def get_identity_client(region=None, profile='DEFAULT'):
     return oci.identity.IdentityClient(get_oci_config(region, profile))
 
 
+def get_object_storage_client(region=None, profile='DEFAULT'):
+    return oci.object_storage.ObjectStorageClient(
+        get_oci_config(region, profile))
+
+
 def service_exception():
     """OCI service exception."""
     return oci.exceptions.ServiceError
+
+
+def with_oci_env(f):
+
+    @functools.wraps(f)
+    def wrapper(*args, **kwargs):
+        # pylint: disable=line-too-long
+        enter_env_cmds = [
+            'conda info --envs | grep "sky-oci-cli-env" || conda create -n sky-oci-cli-env python=3.10 -y',
+            '. $(conda info --base 2> /dev/null)/etc/profile.d/conda.sh > /dev/null 2>&1 || true',
+            'conda activate sky-oci-cli-env', 'pip install oci-cli',
+            'export OCI_CLI_SUPPRESS_FILE_PERMISSIONS_WARNING=True'
+        ]
+        operation_cmd = [f(*args, **kwargs)]
+        leave_env_cmds = ['conda deactivate']
+        return ' && '.join(enter_env_cmds + operation_cmd + leave_env_cmds)
+
+    return wrapper

sky/cloud_stores.py

@@ -7,6 +7,7 @@ TODO:
 * Better interface.
 * Better implementation (e.g., fsspec, smart_open, using each cloud's SDK).
 """
+import os
 import shlex
 import subprocess
 import time
@@ -18,6 +19,7 @@ from sky.adaptors import aws
 from sky.adaptors import azure
 from sky.adaptors import cloudflare
 from sky.adaptors import ibm
+from sky.adaptors import oci
 from sky.clouds import gcp
 from sky.data import data_utils
 from sky.data.data_utils import Rclone
@@ -470,6 +472,64 @@ class IBMCosCloudStorage(CloudStorage):
         return self.make_sync_dir_command(source, destination)
 
 
+class OciCloudStorage(CloudStorage):
+    """OCI Cloud Storage."""
+
+    def is_directory(self, url: str) -> bool:
+        """Returns whether OCI 'url' is a directory.
+        In cloud object stores, a "directory" refers to a regular object whose
+        name is a prefix of other objects.
+        """
+        bucket_name, path = data_utils.split_oci_path(url)
+
+        client = oci.get_object_storage_client()
+        namespace = client.get_namespace(
+            compartment_id=oci.get_oci_config()['tenancy']).data
+
+        objects = client.list_objects(namespace_name=namespace,
+                                      bucket_name=bucket_name,
+                                      prefix=path).data.objects
+
+        if len(objects) == 0:
+            # A directory with few or no items
+            return True
+
+        if len(objects) > 1:
+            # A directory with more than 1 items
+            return True
+
+        object_name = objects[0].name
+        if path.endswith(object_name):
+            # An object path
+            return False
+
+        # A directory with only 1 item
+        return True
+
+    @oci.with_oci_env
+    def make_sync_dir_command(self, source: str, destination: str) -> str:
+        """Downloads using OCI CLI."""
+        bucket_name, path = data_utils.split_oci_path(source)
+
+        download_via_ocicli = (f'oci os object sync --no-follow-symlinks '
+                               f'--bucket-name {bucket_name} '
+                               f'--prefix "{path}" --dest-dir "{destination}"')
+
+        return download_via_ocicli
+
+    @oci.with_oci_env
+    def make_sync_file_command(self, source: str, destination: str) -> str:
+        """Downloads a file using OCI CLI."""
+        bucket_name, path = data_utils.split_oci_path(source)
+        filename = os.path.basename(path)
+        destination = os.path.join(destination, filename)
+
+        download_via_ocicli = (f'oci os object get --bucket-name {bucket_name} '
+                               f'--name "{path}" --file "{destination}"')
+
+        return download_via_ocicli
+
+
 def get_storage_from_path(url: str) -> CloudStorage:
     """Returns a CloudStorage by identifying the scheme:// in a URL."""
     result = urllib.parse.urlsplit(url)
@@ -485,6 +545,7 @@ _REGISTRY = {
     's3': S3CloudStorage(),
     'r2': R2CloudStorage(),
     'cos': IBMCosCloudStorage(),
+    'oci': OciCloudStorage(),
     # TODO: This is a hack, as Azure URL starts with https://, we should
     # refactor the registry to be able to take regex, so that Azure blob can
     # be identified with `https://(.*?)\.blob\.core\.windows\.net`

sky/data/data_transfer.py

@@ -200,3 +200,40 @@ def _add_bucket_iam_member(bucket_name: str, role: str, member: str) -> None:
     bucket.set_iam_policy(policy)
 
     logger.debug(f'Added {member} with role {role} to {bucket_name}.')
+
+
+def s3_to_oci(s3_bucket_name: str, oci_bucket_name: str) -> None:
+    """Creates a one-time transfer from Amazon S3 to OCI Object Storage.
+    Args:
+      s3_bucket_name: str; Name of the Amazon S3 Bucket
+      oci_bucket_name: str; Name of the OCI Bucket
+    """
+    # TODO(HysunHe): Implement sync with other clouds (s3, gs)
+    raise NotImplementedError('Moving data directly from S3 to OCI bucket '
+                              'is currently not supported. Please specify '
+                              'a local source for the storage object.')
+
+
+def gcs_to_oci(gs_bucket_name: str, oci_bucket_name: str) -> None:
+    """Creates a one-time transfer from Google Cloud Storage to
+    OCI Object Storage.
+    Args:
+      gs_bucket_name: str; Name of the Google Cloud Storage Bucket
+      oci_bucket_name: str; Name of the OCI Bucket
+    """
+    # TODO(HysunHe): Implement sync with other clouds (s3, gs)
+    raise NotImplementedError('Moving data directly from GCS to OCI bucket '
+                              'is currently not supported. Please specify '
+                              'a local source for the storage object.')
+
+
+def r2_to_oci(r2_bucket_name: str, oci_bucket_name: str) -> None:
+    """Creates a one-time transfer from Cloudflare R2 to OCI Bucket.
+    Args:
+      r2_bucket_name: str; Name of the Cloudflare R2 Bucket
+      oci_bucket_name: str; Name of the OCI Bucket
+    """
+    raise NotImplementedError(
+        'Moving data directly from Cloudflare R2 to OCI '
+        'bucket is currently not supported. Please specify '
+        'a local source for the storage object.')

sky/data/data_utils.py

@@ -730,3 +730,14 @@ class Rclone():
                 lines_to_keep.append(line)
 
         return lines_to_keep
+
+
+def split_oci_path(oci_path: str) -> Tuple[str, str]:
+    """Splits OCI Path into Bucket name and Relative Path to Bucket
+    Args:
+      oci_path: str; OCI Path, e.g. oci://imagenet/train/
+    """
+    path_parts = oci_path.replace('oci://', '').split('/')
+    bucket = path_parts.pop(0)
+    key = '/'.join(path_parts)
+    return bucket, key

sky/data/mounting_utils.py

@@ -19,6 +19,7 @@ BLOBFUSE2_VERSION = '2.2.0'
 _BLOBFUSE_CACHE_ROOT_DIR = '~/.sky/blobfuse2_cache'
 _BLOBFUSE_CACHE_DIR = ('~/.sky/blobfuse2_cache/'
                        '{storage_account_name}_{container_name}')
+RCLONE_VERSION = 'v1.68.2'
 
 
 def get_s3_mount_install_cmd() -> str:
@@ -30,12 +31,19 @@ def get_s3_mount_install_cmd() -> str:
     return install_cmd
 
 
-def get_s3_mount_cmd(bucket_name: str, mount_path: str) -> str:
+# pylint: disable=invalid-name
+def get_s3_mount_cmd(bucket_name: str,
+                     mount_path: str,
+                     _bucket_sub_path: Optional[str] = None) -> str:
     """Returns a command to mount an S3 bucket using goofys."""
+    if _bucket_sub_path is None:
+        _bucket_sub_path = ''
+    else:
+        _bucket_sub_path = f':{_bucket_sub_path}'
     mount_cmd = ('goofys -o allow_other '
                  f'--stat-cache-ttl {_STAT_CACHE_TTL} '
                  f'--type-cache-ttl {_TYPE_CACHE_TTL} '
-                 f'{bucket_name} {mount_path}')
+                 f'{bucket_name}{_bucket_sub_path} {mount_path}')
     return mount_cmd
 
 
@@ -49,15 +57,20 @@ def get_gcs_mount_install_cmd() -> str:
     return install_cmd
 
 
-def get_gcs_mount_cmd(bucket_name: str, mount_path: str) -> str:
+# pylint: disable=invalid-name
+def get_gcs_mount_cmd(bucket_name: str,
+                      mount_path: str,
+                      _bucket_sub_path: Optional[str] = None) -> str:
     """Returns a command to mount a GCS bucket using gcsfuse."""
-
+    bucket_sub_path_arg = f'--only-dir {_bucket_sub_path} '\
+        if _bucket_sub_path else ''
     mount_cmd = ('gcsfuse -o allow_other '
                  '--implicit-dirs '
                  f'--stat-cache-capacity {_STAT_CACHE_CAPACITY} '
                  f'--stat-cache-ttl {_STAT_CACHE_TTL} '
                  f'--type-cache-ttl {_TYPE_CACHE_TTL} '
                  f'--rename-dir-limit {_RENAME_DIR_LIMIT} '
+                 f'{bucket_sub_path_arg}'
                  f'{bucket_name} {mount_path}')
     return mount_cmd
 
@@ -78,10 +91,12 @@ def get_az_mount_install_cmd() -> str:
     return install_cmd
 
 
+# pylint: disable=invalid-name
 def get_az_mount_cmd(container_name: str,
                      storage_account_name: str,
                      mount_path: str,
-                     storage_account_key: Optional[str] = None) -> str:
+                     storage_account_key: Optional[str] = None,
+                     _bucket_sub_path: Optional[str] = None) -> str:
     """Returns a command to mount an AZ Container using blobfuse2.
 
     Args:
@@ -90,6 +105,7 @@ def get_az_mount_cmd(container_name: str,
             belongs to.
         mount_path: Path where the container will be mounting.
         storage_account_key: Access key for the given storage account.
+        _bucket_sub_path: Sub path of the mounting container.
 
     Returns:
         str: Command used to mount AZ container with blobfuse2.
@@ -106,25 +122,38 @@ def get_az_mount_cmd(container_name: str,
     cache_path = _BLOBFUSE_CACHE_DIR.format(
         storage_account_name=storage_account_name,
         container_name=container_name)
+    if _bucket_sub_path is None:
+        bucket_sub_path_arg = ''
+    else:
+        bucket_sub_path_arg = f'--subdirectory={_bucket_sub_path}/ '
     mount_cmd = (f'AZURE_STORAGE_ACCOUNT={storage_account_name} '
                  f'{key_env_var} '
                  f'blobfuse2 {mount_path} --allow-other --no-symlinks '
                  '-o umask=022 -o default_permissions '
                  f'--tmp-path {cache_path} '
+                 f'{bucket_sub_path_arg}'
                  f'--container-name {container_name}')
     return mount_cmd
 
 
-def get_r2_mount_cmd(r2_credentials_path: str, r2_profile_name: str,
-                     endpoint_url: str, bucket_name: str,
-                     mount_path: str) -> str:
+# pylint: disable=invalid-name
+def get_r2_mount_cmd(r2_credentials_path: str,
+                     r2_profile_name: str,
+                     endpoint_url: str,
+                     bucket_name: str,
+                     mount_path: str,
+                     _bucket_sub_path: Optional[str] = None) -> str:
     """Returns a command to install R2 mount utility goofys."""
+    if _bucket_sub_path is None:
+        _bucket_sub_path = ''
+    else:
+        _bucket_sub_path = f':{_bucket_sub_path}'
     mount_cmd = (f'AWS_SHARED_CREDENTIALS_FILE={r2_credentials_path} '
                  f'AWS_PROFILE={r2_profile_name} goofys -o allow_other '
                  f'--stat-cache-ttl {_STAT_CACHE_TTL} '
                  f'--type-cache-ttl {_TYPE_CACHE_TTL} '
                  f'--endpoint {endpoint_url} '
-                 f'{bucket_name} {mount_path}')
+                 f'{bucket_name}{_bucket_sub_path} {mount_path}')
     return mount_cmd
 
 
@@ -136,9 +165,12 @@ def get_cos_mount_install_cmd() -> str:
     return install_cmd
 
 
-def get_cos_mount_cmd(rclone_config_data: str, rclone_config_path: str,
-                      bucket_rclone_profile: str, bucket_name: str,
-                      mount_path: str) -> str:
+def get_cos_mount_cmd(rclone_config_data: str,
+                      rclone_config_path: str,
+                      bucket_rclone_profile: str,
+                      bucket_name: str,
+                      mount_path: str,
+                      _bucket_sub_path: Optional[str] = None) -> str:
     """Returns a command to mount an IBM COS bucket using rclone."""
     # creates a fusermount soft link on older (<22) Ubuntu systems for
     # rclone's mount utility.
@@ -150,14 +182,60 @@ def get_cos_mount_cmd(rclone_config_data: str, rclone_config_path: str,
                                 'mkdir -p ~/.config/rclone/ && '
                                 f'echo "{rclone_config_data}" >> '
                                 f'{rclone_config_path}')
+    if _bucket_sub_path is None:
+        sub_path_arg = f'{bucket_name}/{_bucket_sub_path}'
+    else:
+        sub_path_arg = f'/{bucket_name}'
     # --daemon will keep the mounting process running in the background.
     mount_cmd = (f'{configure_rclone_profile} && '
                  'rclone mount '
-                 f'{bucket_rclone_profile}:{bucket_name} {mount_path} '
+                 f'{bucket_rclone_profile}:{sub_path_arg} {mount_path} '
                  '--daemon')
     return mount_cmd
 
 
+def get_rclone_install_cmd() -> str:
+    """ RClone installation for both apt-get and rpm.
+    This would be common command.
+    """
+    # pylint: disable=line-too-long
+    install_cmd = (
+        f'(which dpkg > /dev/null 2>&1 && (which rclone > /dev/null || (cd ~ > /dev/null'
+        f' && curl -O https://downloads.rclone.org/{RCLONE_VERSION}/rclone-{RCLONE_VERSION}-linux-amd64.deb'
+        f' && sudo dpkg -i rclone-{RCLONE_VERSION}-linux-amd64.deb'
+        f' && rm -f rclone-{RCLONE_VERSION}-linux-amd64.deb)))'
+        f' || (which rclone > /dev/null || (cd ~ > /dev/null'
+        f' && curl -O https://downloads.rclone.org/{RCLONE_VERSION}/rclone-{RCLONE_VERSION}-linux-amd64.rpm'
+        f' && sudo yum --nogpgcheck install rclone-{RCLONE_VERSION}-linux-amd64.rpm -y'
+        f' && rm -f rclone-{RCLONE_VERSION}-linux-amd64.rpm))')
+    return install_cmd
+
+
+def get_oci_mount_cmd(mount_path: str, store_name: str, region: str,
+                      namespace: str, compartment: str, config_file: str,
+                      config_profile: str) -> str:
+    """ OCI specific RClone mount command for oci object storage. """
+    # pylint: disable=line-too-long
+    mount_cmd = (
+        f'sudo chown -R `whoami` {mount_path}'
+        f' && rclone config create oos_{store_name} oracleobjectstorage'
+        f' provider user_principal_auth namespace {namespace}'
+        f' compartment {compartment} region {region}'
+        f' oci-config-file {config_file}'
+        f' oci-config-profile {config_profile}'
+        f' && sed -i "s/oci-config-file/config_file/g;'
+        f' s/oci-config-profile/config_profile/g" ~/.config/rclone/rclone.conf'
+        f' && ([ ! -f /bin/fusermount3 ] && sudo ln -s /bin/fusermount /bin/fusermount3 || true)'
+        f' && (grep -q {mount_path} /proc/mounts || rclone mount oos_{store_name}:{store_name} {mount_path} --daemon --allow-non-empty)'
+    )
+    return mount_cmd
+
+
+def get_rclone_version_check_cmd() -> str:
+    """ RClone version check. This would be common command. """
+    return f'rclone --version | grep -q {RCLONE_VERSION}'
+
+
 def _get_mount_binary(mount_cmd: str) -> str:
     """Returns mounting binary in string given as the mount command.
 
@@ -209,7 +287,7 @@ def get_mounting_script(
     script = textwrap.dedent(f"""
         #!/usr/bin/env bash
         set -e
-                             
+
         {command_runner.ALIAS_SUDO_TO_EMPTY_FOR_ROOT_CMD}
 
         MOUNT_PATH={mount_path}