Method and system for ensuring cache coherence of metadata in clustered file systems

• US 8,819,357 B2
• Filed: 07/12/2011
• Issued: 08/26/2014
• Est. Priority Date: 07/12/2011
• Status: Active Grant

• Current Assignee: Vmware LLC (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: Vilayannur, Murali, Li, Jinyuan, Vaghani, Satyam B.
• Primary Examiner(s): Rojas, Midys
• Application Number:
  US13/181,129
• Publication Number:
  US 20130019067A1
• Time in Patent Office: 1,141 Days
• Field of Search: 711/130, 711/141, 711/144, 711/145, 711/147, 711/154, 711/156
• US Class Current: 711/156
• CPC Class Codes: G06F 12/08 in hierarchically structured memory systems, e.g. virtual memory systems
• CPC Class Codes: G06F 12/0815 Cache consistency protocols
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache
• CPC Class Codes: G06F 16/1774 Locking methods, e.g. locking methods for file systems allowing shared and concurrent access to files
• CPC Class Codes: G06F 2212/466 Metadata, control data

Abstract

Metadata of a shared file in a clustered file system is changed in a way that ensures cache coherence amongst servers that can simultaneously access the shared file. Before a server changes the metadata of the shared file, it waits until no other server is attempting to access the shared file, and all I/O operations to the shared file are blocked. After writing the metadata changes to the shared file, local caches of the other servers are updated, as needed, and I/O operations to the shared file are unblocked.

Claims

1. A method of ensuring metadata cache coherence of a shared file that is simultaneously accessed by multiple hosts including a first host and one or more other hosts, the first host modifying a metadata of the shared file, the shared file being associated with a lock having a data field for storing a version number, comprising the steps of:

comparing, by the first host, the version number stored in the data field of the lock with a version number that is locally cached within the first host;

if the version numbers match, incrementing the version number and storing the incremented version number in the data field of the lock; and

after said storing, modifying the metadata of the shared file, incrementing the version number once more in response to said modifying, and storing the incremented version number in the data field of the lock.

2. The method of claim 1, wherein after said storing, waiting for a predetermined amount of time before modifying the metadata of the shared file.

3. The method of claim 2, wherein each of the other hosts is configured to carry out the step of comparing once per a time period, and the predetermined amount of time equals n multiplied by a length of the time period, where n is an integer greater than 1.

4. The method of claim 2, wherein each of the other hosts is configured to block I/O operations to the shared file upon detecting that the version number stored in the data field of the lock does not match a locally cached version number.

5. The method of claim 2, wherein each of the other hosts is configured to unblock I/O operations to the shared file upon detecting that the version number stored in the data field of the lock is two greater than a locally cached version number.

6. The method of claim 5, wherein each of the other hosts is further configured to update the metadata of the shared file cached therein upon detecting that the version number stored in the data field of the lock is two greater than a locally cached version number.

7. The method of claim 1, wherein said comparing is performed using a compare-and-write command.

8. A method of ensuring metadata cache coherence of a shared file that is simultaneously accessed by multiple hosts, comprising the steps of:

storing a first value in a predetermined storage location to indicate that the metadata of the shared file is being modified;

determining from a counter value whether or not all other hosts that are accessing the shared file recognize that the metadata of the shared file is being modified, wherein the counter value is initialized to be equal to the number of hosts that are simultaneously accessing the shared file, and the counter value is decremented each time one of the hosts recognizes that the metadata of the shared file is being modified; and

upon determining that all other hosts that are accessing the shared file recognize that the metadata of the shared file is being modified, modifying the metadata of the shared file and storing a second value in the predetermined storage location to indicate that the metadata of the shared file is no longer being modified.

9. The method of claim 8, wherein the predetermined storage location is a logical block number associated with a lock of the shared file.

10. The method of claim 8, wherein each of the hosts, other than a host that is carrying out the metadata modification, is configured to carry out the step of determining once per a first time period, whether or not a value stored in the predetermined storage location is equal to the second value.

11. The method of claim 10, wherein each of the hosts, other than a host that is carrying out the metadata modification, is configured to block I/O operations to the shared file upon detecting that the value stored in the predetermined storage location is not equal to the second value.

12. The method of claim 10, wherein each of the hosts, other than a host that is carrying out the metadata modification, is configured to block I/O operations to decrement the counter value upon detecting that the value stored in the predetermined storage location is not equal to the second value.

13. The method of claim 10, wherein after a second time period equal to n multiplied by the first time period has elapsed after said modifying, where n is an integer greater than 1, determining if an additional metadata modification to the shared file is permitted.

14. The method of claim 8, wherein each of the hosts, other than a host that is carrying out the metadata modification, is configured to unblock I/O operations to the shared file upon detecting that the value stored in the predetermined storage location is equal to the second value.

15. The method of claim 14, wherein each of the hosts, other than a host that is carrying out the metadata modification, is further configured to update the metadata of the shared file cached therein upon detecting that the value stored in the predetermined storage location is equal to the second value.

16. A method of ensuring metadata cache coherence of shared files, each of which is simultaneously accessed by multiple hosts, comprising the steps of:

designating a first one of the multiple hosts as a lock manager for a first shared file; and

designating a second one of the multiple hosts as a lock manager for a second shared file,

wherein the lock manager for a shared file grants a write lock to a modifier host to permit the modifier host to change metadata of the shared file and notifies other hosts that are accessing the shared file of the write lock grant.

17. The method of claim 16, wherein each of the other hosts is configured to block I/O operations to the shared file upon receiving notification of the write lock grant to the modifier host.

18. The method of claim 17, wherein the lock manager for the shared file notifies the other hosts that are accessing the shared file that the metadata changes have been completed.

19. The method of claim 18, wherein each of the other hosts is configured to unblock I/O operations to the shared file and update the metadata of the shared data cached therein upon receiving notification that the metadata changes have been completed.