Systems and methods for storage error management

• US 9,946,607 B2
• Filed: 10/08/2015
• Issued: 04/17/2018
• Est. Priority Date: 03/04/2015
• Status: Active Grant

• Current Assignee: Jpmorgan Chase Bank NA As The Agent
• Original Assignee: SanDisk Technologies LLC (Western Digital Corporation)
• Inventors: Sundararaman, Swaminathan, Talagala, Nisha, Sarto, Eivind, Li, Shaohua
• Primary Examiner(s): Guyton, Philip
• Application Number:
  US14/878,635
• Publication Number:
  US 20160259693A1
• Time in Patent Office: 922 Days
• CPC Class Codes: G06F 11/1088 Reconstruction on already foreseen single or plurality of spare disks
• CPC Class Codes: G06F 11/1458 Management of the backup or restore process
• CPC Class Codes: G06F 16/2365 Ensuring data consistency and integrity
• CPC Class Codes: G06F 2201/805 Real-time

Abstract

A storage layer stores integrity data corresponding to data groups being written to a storage array. The storage layer validates stored data groups by use of the integrity data, independent of recovery information managed by the storage array. The storage layer preempts crash recovery operations of the storage array by validating a subset of the data groups. The storage layer identifies data groups to validate based on a location of a log append point within a storage address space of the storage array. The storage layer may be further configured to mitigate write hole conditions by, inter alia, reconstructing invalid data of a data group and/or retaining valid portions of the data group.

Claims

1. An apparatus, comprising:

a storage service layer configured to;

generate integrity metadata corresponding to data being stored within respective data groups by a storage engine, wherein the storage engine is configured to store the data groups with reconstruction metadata on a storage array comprising a plurality of non-volatile storage devices; and

recover from an invalid shutdown of the storage array by use of the integrity metadata, wherein recovering from the invalid shutdown comprises preventing implementation of a recovery operation of the storage engine pertaining to the invalid shutdown.

2. The apparatus of claim 1, wherein the storage service layer is further configured to determine, by use of the integrity metadata, whether the invalid shutdown resulted in storage of an incomplete data group on the storage array.

3. The apparatus of claim 2, wherein the storage service layer is further configured to identify an incomplete data group stored on the storage array in response to a mismatch between integrity metadata generated for the data group and integrity metadata derived from data read from the storage array.

4. The apparatus of claim 2, wherein the storage service layer is further configured to invalidate at least a portion of a particular data group stored on the storage array in response to determining that the particular data group is incomplete.

5. The apparatus of claim 1, wherein:

the recovery operation of the storage engine comprises validating a set of stored data groups by use of the reconstruction metadata stored with the respective data groups on the storage array; and

the storage service layer is further configured to validate a subset of the set of stored data groups in response to the invalid shutdown.

6. The apparatus of claim 1, wherein the storage service layer is further configured to:

write data at an append point of a storage log on the storage array by use of the storage engine; and

validate a data group corresponding to the append point of the storage log in response to the invalid shutdown of the storage array.

7. The apparatus of claim 1, further comprising a coordination module configured to direct the storage engine to delegate recovery from the invalid shutdown to the storage service layer.

8. The apparatus of claim 1, further comprising a storage array coordinator configured to prevent implementation of a resynchronization operation by the storage engine in response to the invalid shutdown, the resynchronization operation to validate data groups stored on the storage array by use of the reconstruction metadata stored with the data groups by the storage engine.

9. A system, comprising:

a storage layer configured to store integrity data corresponding to data being stored within respective data stripes of a storage system, wherein the data stripes stored within the storage system correspond to respective parity reconstruction data generated for the data stripes by the storage system;

a crash recovery module configured to validate one or more data stripes of the storage system by use of the stored integrity data in response to an invalid shutdown of the storage system; and

a storage coordination module configured to block performance of a recovery operation by the storage system in response to the invalid shutdown.

10. The system of claim 9, wherein the crash recovery module is further configured to detect the invalid shutdown of the storage system.

11. The system of claim 9, wherein:

the storage system comprises a Redundant Array of Inexpensive Disks (RAID) storage system; and

the storage coordination module is further configured to cause the RAID storage system to delegate recovery operations pertaining to invalid shutdowns of the storage system to the storage layer.

12. The system of claim 9, wherein:

the crash recovery module is configured to validate a particular data stripe in response to the invalid shutdown; and

validating the particular data stripe comprises determining whether the particular data stripe comprises a write error by comparing stored integrity data of the particular data stripe to integrity data generated from the particular data stripe stored within the storage system.

13. The system of claim 9, wherein:

the storage coordination module is further configured to transmit a message to the RAID storage system in response to the invalid shutdown, the message configured to block performance of the recovery operation by the storage system, the recovery operation comprising a Redundant Array of Inexpensive Disks (RAID) resynchronization operation.

14. The system of claim 9, wherein:

a particular data stripe stored within the storage system comprises a plurality of data units;

the integrity data stored for the particular data stripe comprises a plurality of integrity datum, each integrity datum corresponding to a respective data unit of the plurality of data units; and

the crash recovery module is further configured to validate respective data units of the particular data stripe stored within the storage system in response to the invalid shutdown, by comparing the integrity datum stored for the particular data stripe with integrity datum derived from the respective data units of the particular data stripe stored within the storage system.

15. A non-transitory computer-readable storage medium storing program code configured to cause a computing device to perform operations, the operations comprising:

appending data groups to a storage log maintained on a storage array by use of a storage array controller, the storage array controller configured to store the data groups within respective data stripes on the storage array, wherein appending the data groups further comprises storing checksum values corresponding to the data groups; and

in response to an invalid shutdown of the storage array, the operations further comprise;

validating an identified data stripe stored at a head of the storage log on the storage array by use of the stored checksum values corresponding to the data groups; and

preventing execution of a crash recovery operation by the storage array controller.

16. The non-transitory computer-readable storage medium of claim 15, the operations further comprising:

identifying the data stripe at the head of the storage log in response to the invalid shutdown; and

comparing a checksum value derived from the identified data stripe to a stored checksum value corresponding to the identified data stripe.

17. The non-transitory computer-readable storage medium of claim 16, the operations further comprising:

invalidating the identified data stripe in response to a mismatch between the checksum value derived from the identified data stripe and the stored checksum value corresponding to the identified data stripe.

18. The non-transitory computer-readable storage medium of claim 16, wherein:

the data stripe at the head of the storage log comprises a plurality of data blocks mapped to respective addresses of a logical address space; and

the operations further comprise invalidating associations between the data stripe and the respective addresses in response to a mismatch between the checksum value derived from the identified data stripe and the stored checksum value corresponding to the identified data stripe.

19. The non-transitory computer-readable storage medium of claim 15, wherein:

storing the checksum values comprises appending mapping entries comprising the checksum values to a metadata log maintained on a storage device that is independent of the storage array, and

the mapping entries associate data blocks within the respective data stripes with respective logical addresses of a logical address space.

20. The non-transitory computer-readable storage medium of claim 15, wherein storing a checksum value of a data stripe comprises including the checksum value in the data group stored within the data stripe on the storage array.

21. The non-transitory computer-readable storage medium of claim 15, wherein:

the stored checksum values for a data group stored within the identified data stripe comprise a plurality of segment checksum values, each segment checksum value corresponding to a respective data segment of the data group; and

validating the identified data stripe further comprises;

comparing respective checksum values stored for each of a plurality of data segments of a data group to checksum values derived from corresponding data segments stored within the identified data stripe;

invalidating a first data segment stored within the identified data stripe in response to a checksum mismatch pertaining to the first data segment; and

rewriting a second data segment stored within the identified data stripe to a different data stripe on the storage array in response to a checksum match pertaining to the second data segment.