Metadata offload for a file server cluster

• US 20070055702A1
• Filed: 09/07/2005
• Published: 03/08/2007
• Est. Priority Date: 09/07/2005
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Faibish, Sorin, Jiang, Xiaoye, Fridella, Stephen, Forecast, John
• Granted Patent: US 7,617,216 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/122 using management policies backup systems G06F11/1446; file migration policies for HSM systems G06F16/185
• CPC Class Codes: G06F 16/182 Distributed file systems

Abstract

A file server system has a cluster of server computers that share access to a file system in shared storage. One of the server computers has primary responsibility for management of access to the file system. In order to reduce the possibility of primary server overload when a large number of the clients happen to concurrently access the same file system, most metadata processing operations are offloaded to secondary server computers. This also facilitates recovery from failure of a primary server computer since only a fraction of the ongoing metadata operations of a primary server computer is interrupted by a failure of the primary server computer. For example, a secondary data mover may truncate, delete, create, or rename a file in response to a client request.

Claims

1. In a file server system having server computers and storage shared among the server computers, the shared storage containing a file system, access to the file system being managed by one of the server computers that is primary with respect to the file system, other server computers being secondary with respect to the file system, a method comprising:

one of the server computers that is secondary with respect to the file system responding to a request from a client for creating, deleting, or renaming a file in the file system by requesting an exclusive lock on a source or target directory containing the file from the server computer that is primary with respect to the file system, and

the server computer that is primary with respect to the file system responding to said one of the server computers that is secondary with respect to the file system by the server computer that is primary with respect to the file system granting an exclusive lock on the source or target directory and returning a copy of an inode of the source or target directory to said one of the server computers that is secondary with respect to the file system; and

said one of the server computers that is secondary with respect to the file system modifying the inode of the source or target directory in order to create, delete, or rename the file in the file system.

2. The method as claimed in claim 1, wherein said one of the server computers that is secondary with respect to the file system has a log file for the file system, and the method further includes said one of the server computers that is the secondary with respect to the file system logging the modification to the inode of the source or target directory to the log file for the file system.

3. The method as claimed in claim 1, wherein said one of the server computers that is secondary with respect to the file system modifies the inode of the source or target directory by sending a request to the server computer that is primary with respect to the file system for modifying the inode of the source or target directory.

4. The method as claimed in claim 3, which includes the server computer that is primary with respect to the file system responding to the request for modifying the inode of the source or target directory by writing to the inode of the source or target directory in an inode cache of the server computer that is primary with respect to the file system, and at a later time, flushing the inode of the source or target directory from the inode cache of the server computer that is primary with respect to the file system to disk storage.

5. The method as claimed in claim 1, which includes said one of the server computers that is secondary with respect to the file system creating the file in the target directory by moving an inode for the file to the target directory from an allocation directory for said one of the server computers that is secondary with respect to the file system.

6. The method as claimed in claim 1, which includes said one of the server computers that is secondary with respect to the file system extending the allocation directory by requesting the server computer that is primary with respect to the file system to add inodes to the allocation directory for said one of the server computers that is secondary with respect to the file server, and the server computer that is primary with respect to the file system adding inodes to the allocation directory for said one of the server computers that is secondary with respect to the file server.

7. The method as claimed in claim 1, which includes said one of the server computers that is secondary with respect to the file system deleting the file from the source directory by moving an inode for the file from the source directory to an allocation directory for said one of the server computers that is secondary with respect to the file system.

8. The method as claimed in claim 1, which includes said one of the server computers that is secondary with respect to the file system renaming the file by moving an inode for file from the source directory to the target directory.

9. The method as claimed in claim 1, which includes recovering from a crash of said one of the server computers that is secondary with respect to the file system by another one of the server computers replaying a log of said one of the server computers that is secondary with respect to the file system.

10. The method as claimed in claim 1, which includes recovering from a crash of the server computer that is primary with respect to the file system by selecting another server computer to become primary with respect to the file system, and said another server computer replaying a log of the crashed server computer, said another server computer requesting said one of the server computers that is secondary with respect to the file system to flush a log of said one of the server computers that is secondary with respect to the file system.

11. A file server system comprising shared storage, and multiple server computers linked to the shared storage for accessing a file system in the shared storage, wherein one of the server computers is primary with respect to the file system for managing access to the file system, and others of the server computers are secondary with respect to the file system,

wherein each of the server computers that are secondary with respect to the file system is programmed for responding to a client request for creating, deleting, or renaming a file in the file system by requesting an exclusive lock on a source or target directory for containing the file from the server computer that is primary with respect to the file system, and

wherein the server computer that is primary with respect to the file system is programmed for responding to said each of the server computers that are secondary with respect to the file system by the server computer that is primary with respect to the file system granting an exclusive lock on the source or target directory and returning a copy of an inode of the source or target directory to said each of the server computers that are secondary with respect to the file system;

and

said each of the server computers that are secondary with respect to the file system is programmed for modifying the inode of the source or target directory in order to create, delete, or rename the file in the file system.

12. The system as claimed in claim 11, wherein said each of the server computers that are secondary with respect to the file system has a respective log file for the file system, and said each of the server computers that is the secondary with respect to the file system is programmed for logging the modification to the inode of the source or target directory to the respective log file for the file system.

13. The system as claimed in claim 11, wherein said each of the server computers that are secondary with respect to the file system is programmed for modifying the inode of the source or target directory by sending a request to the primary server computer for modifying the inode of the source or target directory.

14. A file server system comprising shared storage, and multiple server computers linked to the shared storage for accessing a file system in the shared storage, wherein one of the server computers is primary with respect to the file system for managing access to the file system, and others of the server computers are secondary with respect to the file system,

wherein the server computer that is primary with respect to the file system is programmed for responding to a respective request from each of the server computers that are secondary with respect to the file system for requesting the server computer that is primary with respect to the file system to grant a read lock and return directory mapping information for a specified directory;

wherein the server computer that is primary with respect to the file system is programmed for responding to a respective request from said each of the server computers that are secondary with respect to the file system for requesting the server computer that is primary with respect to the file system to extend a specified file;

wherein the server computer that is primary with respect to the file system is programmed for responding to a respective request from said each of the server computers that are secondary with respect to the file system for requesting the server computer that is primary with respect to the file system to commit a specified file to disk storage;

wherein the server computer that is primary with respect to the file system is programmed for responding to a respective request from said each of the server computers that are secondary with respect to the file system for requesting the server computer that is primary with respect to the file system to grant an exclusive lock on an inode for a specified file and return a copy of the inode for the specified file; and

wherein the server computer that is primary with respect to the file system is programmed for responding to a respective request from said each of the server computers that are secondary with respect to the file system for requesting the server computer that is primary with respect to the file system to write changes to an inode for a specified file.

15. The file server system as claimed in claim 14, wherein the server computer that is primary with respect to the file system is programmed for responding to a respective request from each of the server computers that are secondary with respect to the file system for extending an allocation directory for said each of the server computers that are secondary with respect to the file system by adding inodes to the allocation directory for said each of the server computers that are secondary with respect to the file system.

16. The file server system as claimed in claim 14, which is programmed for recovering from a crash of the server computer that is primary with respect to the file system by at least one of the server computers replaying a log of the server computer that is primary with respect to the file system, and said each of the server computers that are secondary with respect to the file system flushing metadata changes for the file system to the shared storage.

17. The file server system as claimed in claim 14, which is programmed for recovering from a crash of one of the server computers that are secondary with respect to the file system by another of the server computers replaying a log of said one of the server computers that are secondary with respect to the file system.

18. The file server system as claimed in claim 14, wherein the file system includes an inode pool, a data block pool, an inode bit map, a data block bit map, and

for said each of the server computers that are secondary with respect to the file system:

a respective allocation directory for providing inodes to be used by said each of the server computers that are secondary with respect to the file system for adding inodes to directories; and

a respective allocation file for providing file blocks to be used by said each of the server computers that are secondary with respect to the file system for adding file blocks to files.

19. The file server system as claimed in claim 14, wherein:

the file system includes a metadata log; and

the metadata log includes a respective log file for the server computer that is primary with respect to the file system, and a respective log file for each of the server computers that are secondary with respect to the file system.

20. The file server system as claimed in claim 14, wherein said each of the server computers that are secondary with respect to the file system is programmed for updating, in a local inode cache, file access time with respect to a respective file in the file system without logging a metadata change, and later writing the respective file access time update to the file system in the shared storage upon release of access rights to the respective file in said one of the file systems.