Abstract: A server node can monitor the status of servers in a server cluster. The node may receive an alert indicating that a server in the server cluster is unavailable. In response to the alert, the node can send instructions that cause pending remote procedure call requests to be canceled and then reissued to another server in the server cluster instead of to the first server.

Abstract: A system and method for replicating a set of files from a source computer system to a target computer system are described. The underlying extent structure of the files may be preserved. For example, if an original file on the source computer system has a particular number of extents then the copy of the file created on the target computer system may have the same number of extents, where each respective extent in the copy represents the same byte range of the file as its corresponding extent in the original file. The file systems employed on the source computer system and the target computer system may support different extent types. The extent types of the original files stored on the source computer system may be preserved in the copies of the files on the target computer system.

Abstract: A lock is acquired on a data structure. Content in the data structure is read and verified while the lock is held. The lock is then released, and then the file system components that are referred to by the data structure are verified. In essence, a file system consistency check of the file system components is performed offline in the background while the data structure remains accessible.

Abstract: A list of data structures (e.g., inodes) can be accessed, and the data structures in the list can be examined. If a data structure is examined, a counter value associated with the data structure is changed to a generation number that is associated with the examination. Subsequently, the counter values can be used to identify unreferenced data structures. More specifically, the counter value for an unreferenced data structure will be different from the generation number for the most recently performed examination.

Abstract: A lock is acquired on a data structure. Content in the data structure is read and verified while the lock is held. The lock is then released, and then the file system components that are referred to by the data structure are verified. In essence, a file system consistency check of the file system components is performed offline in the background while the data structure remains accessible.

Abstract: A list of data structures (e.g., inodes) can be accessed, and the data structures in the list can be examined. If a data structure is examined, a counter value associated with the data structure is changed to a generation number that is associated with the examination. Subsequently, the counter values can be used to identify unreferenced data structures. More specifically, the counter value for an unreferenced data structure will be different from the generation number for the most recently performed examination.

Abstract: A server node can monitor the status of servers in a server cluster. The node may receive an alert indicating that a server in the server cluster is unavailable. In response to the alert, the node can send instructions that cause pending remote procedure call requests to be canceled and then reissued to another server in the server cluster instead of to the first server.

Abstract: Embodiments of a system and method for distributed file system I/O recovery in storage networks. Embodiments may detect loss of access to a server in the storage network and recover application I/O requests in real-time once access to the server is restored. Embodiments may detect server and/or network failures and store failed and new I/O requests. Recovery from the failure (e.g. network reconnect, server node reboot, or failover, if this is a clustered environment) may be detected and, after recovery is detected, any stored failed and new I/O requests may be sent to the server. In one embodiment, to detect recovery from the failure, a failed I/O request may be repeatedly re-issued until the I/O request succeeds. Embodiments may be implemented in a variety of storage environments, including environments where clients issue direct I/O over a storage network to storage and control I/O over a network to a server.