Locating previous versions of an object in a storage cluster

摘要

Technology for accessing previous versions of data objects within a storage cluster including multiple storage volumes under a common namespace is disclosed herein. In some embodiments, the network storage cluster having multiple cooperative storage nodes receives a namespace identifier for a previous version of a data object. The namespace identifier includes a namespace path containing a volume junction and a time indicator indicating a point in time when the previous version of the data object existed. The network storage cluster identifies a storage volume within the network storage cluster using the volume junction. The volume junction points to a root directory of a file system of the storage volume. The network storage cluster locates, based on the namespace path, the previous version of the data object within a snapshot copy of the storage volume, wherein the snapshot copy is identified by the time indicator.

主权项

1. A method, comprising:
receiving, at a network storage cluster that includes multiple cooperative storage nodes and multiple storage volumes, a namespace identifier for a previous version of a data object, wherein the namespace identifier includes a namespace path containing a volume junction and a time indicator indicating a point in time when the previous version of the data object existed, wherein a first storage volume at a first node contains the identified volume junction as a sub-directory within a first root directory of a first file system of the first storage volume, wherein the volume junction points to a second root directory of a second file system of a second storage volume at a second node; identifying, by a processor, the second storage volume within the network storage cluster by using the volume junction identified by the namespace identifier, wherein identifying the second storage volume within the network storage cluster by using the volume junction identified by the namespace identifier comprises:
determining, by the first file system, that the namespace path to the first file system of the first node includes a first portion identifying a first path with multiple sub-directories in the first file system and that the path includes the volume junction followed by a second portion identifying a second path with multiple sub-directories that are not in the first file system,determining, by the second file system, that the second path includes multiple sub-directories in the second file system, andselecting, by a namespace manager of the network storage cluster, the second storage volume based on the second portion of the namespace path lacking any volume junctions and the second file system manager determining that the second path includes multiple sub-directories in the second file system; and locating, by the processor based on the second volume identified from the volume junction in the namespace path, the previous version of the data object within a snapshot copy of the second storage volume, wherein the snapshot copy is identified using the time indicator in the namespace identifier.