Abstract: A system can receive a file in parts. The system can write a first part to a first temporary file that corresponds to the first part's length at a first location of the first temporary file based on the its part number. The system can for a second part having a same length as the first part, write the second part to the first temporary file at a second location of the first temporary file based on its part number. The system can, in response to receiving a third part that has a different length than the first part, write the third part to a second temporary file at a third location of the second temporary file based on its part number. The system can, after receiving the parts, write data of the second temporary file to an end of the first temporary file, and rename the temporary file.

Abstract: Described herein are technologies directed to multipart upload. A cluster coherent unique identifier for a multipart upload can be generated by creating a leaf in a B-tree. The leaf in the B-tree can comprise a key, and the key can comprise an upload identifier and a group identifier associated with a group of multipart uploads. A parts directory can be created for the multipart upload, wherein elements of the parts directory can be identified using the group identifier and the upload identifier. Upload parts can be transmitted from a client device to a server device, and stored in the parts directory. To complete the multipart upload, the upload parts can be concatenated and stored at a target location.

Abstract: Techniques are provided for consistent entity tags with multiple protocol data access. In an example, a file storage system is configured to process data according to file storage protocol(s) and object storage protocol(s). An object storage protocol can utilize entity tags that indicate whether an object (represented with a file in the file storage system) has changed. Where a file storage protocol is utilized to modify a file, an indication may be stored that indicates that the file lacks a valid entity tag. If an object storage operation is made to retrieve an object, and if the object corresponds to a valid entity tag, then that entity tag can be returned as part of the response. If the object does not correspond to a valid entity tag, then the file storage system can generate a new entity tag and return the newly generated entity tag as part of the response.

Abstract: A system can receive a file in parts. The system can write a first part to a first temporary file that corresponds to the first part's length at a first location of the first temporary file based on the its part number. The system can for a second part having a same length as the first part, write the second part to the first temporary file at a second location of the first temporary file based on its part number. The system can, in response to receiving a third part that has a different length than the first part, write the third part to a second temporary file at a third location of the second temporary file based on its part number. The system can, after receiving the parts, write data of the second temporary file to an end of the first temporary file, and rename the temporary file.

Abstract: Described herein are technologies directed to resumable ordered recursive traversal of an unordered directory tree. Using the techniques described herein, a lexicographic listing of stored objects can be efficiently built from a directory tree that is not lexicographically ordered. Furthermore, the techniques provided herein can output an initial partial lexicographic listing of stored objects having a start location and an end location, and later resumed to output a subsequent lexicographic listing of stored objects which begins after the end location.

Abstract: Described herein are technologies directed to multipart upload. A cluster coherent unique identifier for a multipart upload can be generated by creating a leaf in a B-tree. The leaf in the B-tree can comprise a key, and the key can comprise an upload identifier and a group identifier associated with a group of multipart uploads. A parts directory can be created for the multipart upload, wherein elements of the parts directory can be identified using the group identifier and the upload identifier. Upload parts can be transmitted from a client device to a server device, and stored in the parts directory. To complete the multipart upload, the upload parts can be concatenated and stored at a target location.

Abstract: Techniques are provided for consistent entity tags with multiple protocol data access. In an example, a file storage system is configured to process data according to file storage protocol(s) and object storage protocol(s). An object storage protocol can utilize entity tags that indicate whether an object (represented with a file in the file storage system) has changed. Where a file storage protocol is utilized to modify a file, an indication may be stored that indicates that the file lacks a valid entity tag. If an object storage operation is made to retrieve an object, and if the object corresponds to a valid entity tag, then that entity tag can be returned as part of the response. If the object does not correspond to a valid entity tag, then the file storage system can generate a new entity tag and return the newly generated entity tag as part of the response.

Abstract: Techniques are provided for keeping object access on a file store consistent with other file protocols. In an example, a server that stores data in a file system receives a PUT OBJECT operation to perform, which corresponds to a target file. The server creates a temporary file and writes to the temporary file. After the writing, the server can lock the target file, rename the temporary file to the name of the target file, and unlock the target file. In another example, the server receives a GET OBJECT operation, which corresponds to a target file. The server locks the target file, reads the first file, and releases the lock on the target file. This approach can maintain consistency between object storage operations and file storage operations that are implemented by the server.