Address review comments & more testing.

examples/oci/dataset-upload-and-mount.yaml

@@ -16,6 +16,13 @@ file_mounts:
     persistent: True  # Defaults to True; can be set to false. Optional.
     mode: MOUNT  # Either MOUNT or COPY. Optional.
 
+  /datasets-storage2:
+    name: skybucket2  # Name of storage, optional when source is bucket URI
+    source: './examples/oci'  # Source path, can be local or bucket URL. Optional, do not specify to create an empty bucket.
+    store: oci  # E.g 'oci', 's3', 'gcs'...; default: None. Optional.
+    persistent: True  # Defaults to True; can be set to false. Optional.
+    mode: MOUNT  # Either MOUNT or COPY. Optional.
+
 # Working directory (optional) containing the project codebase.
 # Its contents are synced to ~/sky_workdir/ on the cluster.
 workdir: .
@@ -34,3 +41,7 @@ run: |
   ls -lthr /datasets-storage
   echo "hi" >> /datasets-storage/foo.txt
   ls -lthr /datasets-storage
+
+  ls -lthr /datasets-storage2
+  echo "hi" >> /datasets-storage2/foo2.txt
+  ls -lthr /datasets-storage2

sky/data/storage.py

@@ -3664,6 +3664,9 @@ def _validate(self):
                 assert self.name == data_utils.split_oci_path(self.source)[0], (
                     'OCI Bucket is specified as path, the name should be '
                     'the same as OCI bucket.')
+            elif not re.search(r'^\w+://', self.source):
+                # Treat it as local path.
+                pass
             else:
                 raise NotImplementedError(
                     f'Moving data from {self.source} to OCI is not supported.')
@@ -3826,7 +3829,7 @@ def get_dir_sync_command(src_dir_path, dest_dir_name):
                 'oci os object bulk-upload --no-follow-symlinks --overwrite '
                 f'--bucket-name {self.name} --namespace-name {self.namespace} '
                 f'--object-prefix "{dest_dir_name}" --src-dir "{src_dir_path}" '
-                f'{excludes}" ')
+                f'{excludes} ')
 
             return sync_command