MANAGING CONCURRENT FILE SYSTEM ACCESSES BY MULTIPLE SERVERS USING LOCKS

• US 20110179082A1
• Filed: 03/29/2011
• Published: 07/21/2011
• Est. Priority Date: 02/06/2004
• Status: Abandoned Application

• Current Assignee: VMware, Inc. (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: RAJASHEKHAR, Manjunath, VAGHANI, Satyam B.
• Application Number:
  US13/074,916
• Publication Number:
  US 20110179082A1
• Time in Patent Office: Days
• US Class Current: 707/781
• CPC Class Codes: G06F 16/1774 Locking methods, e.g. locking methods for file systems allowing shared and concurrent access to files

Abstract

Atomic test and set (ATS) operations are carried out to perform lock operations that allow a node to acquire or release a lock to a resource of a shared file system that is stored in a data storage unit (DSU) and update its liveness information. Each ATS operation includes the step of comparing contents accessed and read through the shared file system and contents stored at a particular logical block number of the DSU. If the two contents match, updates to the contents of the lock or the liveness information are permitted.

Claims

1. A method of managing accesses to a resource of a shared file system that is stored in a data storage unit (DSU), comprising:

reading a lock associated with the resource to obtain a current state of the lock;

determining that the lock is available based on the current state;

transmitting a request to the DSU to perform an atomic update to the lock comprising a first operation to confirm that the current state of the lock has not changed since the reading and a second operation to acquire the lock, wherein no other operation can be performed on the lock between the first operation and second operation; and

acquiring access to the resource upon receiving confirmation of successful completion of the atomic update, whereby no exclusive reservation of the DSU is required to acquire the lock.

2. The method of claim 1, wherein the atomic update is a “

compare and write”

SCSI command.

3. The method of claim 1, further comprising receiving an indication that the atomic update has failed if an intervening operation changes the current state of the lock between the reading step and the transmitting step.

4. The method of claim 1, performed by a host computer system coupled to the DSU, wherein the resource is a virtual disk file corresponding to a virtual machine running on the host computer system.

5. The method of claim 1, wherein the lock comprises an owner ID field and a lease field specifying a period of time for possessing the lock.

6. The method of claim 5, wherein the lock is determined to be available if there is no valid owner ID value in the owner ID field.

7. The method of claim 5, wherein the lock is determined to be available if the period of time in the lease field has expired.

8. The method of claim 5, wherein the lock further comprises liveness information indicating whether a host computer system possessing the lock is currently in communication with the DSU.

9. A non-transitory computer-readable storage medium including instructions for managing accesses to a resource of a shared file system that is stored in a data storage unit (DSU), that when executed by a computer processor, perform the steps of:

reading a lock associated with the resource to obtain a current state of the lock;

determining that the lock is available based on the current state;

transmitting a request to the DSU to perform an atomic update to the lock comprising a first operation to confirm that the current state of the lock has not changed since the reading and a second operation to acquire the lock, wherein no other operation can be performed on the lock between the first operation and second operation; and

acquiring access to the resource upon receiving confirmation of successful completion of the atomic update, whereby no exclusive reservation of the DSU is required to acquire the lock.

10. The non-transitory computer-readable storage medium of claim 9, wherein the atomic update is a “

compare and write”

SCSI command.

11. The non-transitory computer-readable storage medium of claim 9, wherein the instructions further comprise receiving an indication that the atomic update has failed if an intervening operation changes the current state of the lock between the reading step and the transmitting step.

12. The non-transitory computer-readable storage medium of claim 9, wherein the instructions are executed by a host computer system coupled to the DSU and wherein the resource is a virtual disk file corresponding to a virtual machine running on the host computer system.

13. The non-transitory computer-readable storage medium, wherein the lock comprises an owner ID field and a lease field specifying a period of time for possessing the lock.

14. A method of updating a heartbeat region associated with a node and stored in a data storage unit (DSU), comprising:

identifying a heartbeat region associated with a node, wherein the heartbeat region stores liveness information associated with the node;

generating updated liveness information associated with the node; and

performing an atomic update operation on the heartbeat region to store the updated liveness information in the heartbeat region, wherein at least one other resource of the shared file system is accessible while the atomic update operation is being performed.

15. The method of claim 14, wherein the atomic update operation is an atomic test and set (ATS) operation.

16. The method of claim 15, wherein the heartbeat region includes a logical block number (LBN) of the DSU at which the heartbeat region is located, and the ATS operation is performed using the LBN.

17. The method of claim 16, further comprising the step of reading the contents of the heartbeat region through a shared file system, wherein the ATS operation includes comparing the contents of the heartbeat region as read through the shared file system and the contents of the heartbeat region stored at the LBN.

18. The method of claim 17, wherein the ATS operation fails when the contents of the heartbeat region as read through the shared file system do not match the contents of the heartbeat region stored at the LBN.

19. The method of claim 14, wherein the updated liveness information includes an updated pulse that indicates that the node is alive.

20. The method of claim 14, wherein the updated liveness information includes an updated heartbeat generation number that is incremented when the heartbeat region is allocated to the node.