Transactions and failure

• US 9,053,024 B2
• Filed: 11/30/2012
• Issued: 06/09/2015
• Est. Priority Date: 11/30/2012
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Graefe, Goetz
• Primary Examiner(s): DUNCAN, MARC M
• Application Number:
  US13/690,353
• Publication Number:
  US 20140157048A1
• Time in Patent Office: 921 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/0745 in an input/output transactions management context input/output processing in general G06F13/00
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 11/1474 in transactions G06F16/20 takes precedence
• CPC Class Codes: G06F 16/2365 Ensuring data consistency and integrity

Abstract

Disclosed herein are techniques for system recovery. It is determined whether a transaction depends on a change committed before a failure. It is further determined whether a transaction conflicts with a change interrupted by the failure.

Claims

1. A system comprising:

a re-do module which, if executed, instructs at least one processor to repeat a change committed by a transaction before a failure but not persisted due to the failure, the change being required by a resumed transaction, to allow the resumed transaction dependent on the change to execute successfully; and

an un-do module which, if executed, instructs at least one processor to revert a change interrupted by the failure during a transaction, the change being required by a resumed transaction, to allow the resumed transaction conflicting with the interrupted change to execute successfully.

2. The system of claim 1, wherein the change committed before the failure comprises an update to a page of data in a buffer.

3. The system of claim 2, wherein the re-do module, if executed, further instructs at least one processor to persist the change in the buffer to a data storage device.

4. The system of claim 1, further comprising a log containing a record of each transaction executed in the system.

5. The system of claim 4, wherein the re-do module, if executed, instructs at least one processor to determine whether the change committed before the failure but did not persist due to the failure based on information contained in the log.

6. The system of claim 4, wherein the un-do module, if executed, instructs at least one processor to determine the change was interrupted by the failure based on information in the log.

7. The system of claim 4, wherein the un-do module, if executed, instructs at least one processor to determine the resumed transaction conflicting with the interrupted change based on information in the log.

8. A non-transitory computer readable medium having instructions therein which, if executed, cause at least one processor to:

determine whether a resumed transaction depends on a change committed before a failure but not persisted due to the failure;

if the resumed transaction depends on the committed change, redo the change and continue execution of the resumed transaction;

determine whether the resumed transaction conflicts with a change interrupted by the failure; and

if the resumed transaction conflicts with the interrupted change, revert the interrupted change and continue execution of the resumed transaction.

9. The non-transitory computer readable medium of claim 8, wherein the committed change comprises an update to a block of data in a buffer.

10. The non-transitory computer readable medium of claim 9, wherein the instructions stored therein, if executed, further cause at least one processor to persist the committed change in the buffer to a storage device.

11. The non-transitory computer readable medium of claim 9, wherein the instructions stored therein, if executed, further cause at least one processor to determine which transactions were scheduled for execution after the failure using a log containing a record of each transaction executed in a system.

12. The non-transitory computer readable medium of claim 11, wherein the instructions stored therein, if executed, further cause at least one processor to obtain information associated with the committed change from the log, when it is determined that the resumed transaction is dependent on the committed change.

13. The non-transitory computer readable medium of claim 11, wherein the instructions stored therein, if executed, further cause at least one processor to obtain information associated with the interrupted change from the log, when it is determined that the resumed transaction conflicts with the interrupted change.

14. A method, comprising:

determining, using at least one processor, whether a resumed transaction depends on a change committed before a failure but not persisted due to failure;

if the resumed transaction depends on the change, repeating, using at least one processor, the committed change to allow the resumed transaction dependent thereon to execute successfully;

determining, using at least one processor, whether the resumed transaction conflicts with a change interrupted by the failure; and

if the resumed transaction conflicts with the interrupted change, reverting, using at least one processor, the interrupted change to allow the resumed transaction conflicting therewith to execute successfully.

15. The method of claim 14, wherein the committed change comprises an update to a block of data in a buffer.

16. The method of claim 14, further comprising obtaining, using at least one processor, information associated with a transaction prior to system failure from a log containing a record of each transaction executed in a system.

17. The method of claim 16, further comprising obtaining, using at least one processor, information associated with the committed change from the log, when it is determined that the resumed transaction depends thereon.

18. The method of claim 17, further comprising obtaining, using at least one processor, information associated with the interrupted change from the log, when it is determined that the resumed transaction conflicts therewith.