Maintaining data integrity in data migration operations using per-migration device error flags

• US 8,977,896 B1
• Filed: 03/29/2013
• Issued: 03/10/2015
• Est. Priority Date: 03/13/2013
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Thigpen, Edward L., Bappe, Michael E.
• Primary Examiner(s): LE, DIEU MINH T
• Application Number:
  US13/853,291
• Time in Patent Office: 711 Days
• Field of Search: 714/15, 714/16, 714/21, 714/3, 714/6.1, 714/6.11, 714/42
• US Class Current: 714/15
• CPC Class Codes: G06F 11/0727 in a storage system, e.g. in a DASD or network based storage system drivers for digital recording or reproducing units G06F3/06; circuits for error detection or correction within digital recording or reproducing units G11B20/18; for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS], H04L67/1097
• CPC Class Codes: G06F 11/0766 Error or fault reporting or storing
• CPC Class Codes: G06F 11/14 Error detection or correction of the data by redundancy in operation G06F11/16 takes precedence
• CPC Class Codes: G06F 11/1441 Resetting or repowering
• CPC Class Codes: G06F 11/1662 the resynchronized component or unit being a persistent storage device re-synchronization of failed mirror storage G06F11/2082; rebuild or reconstruction of parity RAID storage G06F11/1008
• CPC Class Codes: G06F 11/2094 Redundant storage or storage space G06F11/2056 takes precedence
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups
• CPC Class Codes: G06F 3/06 Digital input from, or digital output to, record carriers , e.g. RAID, emulated record carriers or networked record carriers

Abstract

A data migration operation uses an error flag to protect against data corruption. During write cloning states, the error flag is set and only allowed state transitions are performed, including a non-error transition to a committed state when no device fault is recorded for a target storage device and an error transition to a setup state when a device fault is recorded for the target storage device. In a clean system shutdown, a shutdown procedure records a detected target storage device fault and clears the error flag; the recorded device fault later forces the error transition of the migration operation. During the system startup, if the error flag is set then a target storage device fault is recorded to likewise later force the error transition of the migration operation, on the assumption that a detected fault may exist but be unrecorded because the shutdown procedure did not complete during shutdown.

Claims

1. A method of managing state transitions in a data migration operation, the migration operation including an initial setup state, a set of write cloning states in which a target storage device is becoming or is maintained synchronized with a source storage device, and a later committed state in which the target storage device is used to the exclusion of the source storage device, the set of write cloning states spanning a system shutdown/startup cycle, comprising:

upon entering the write cloning states, setting an error flag used to protect against data corruption in the event of a non-clean system shutdown;

during the write cloning states, performing only allowed state transitions based on recorded device faults, the allowed state transitions including a non-error transition to the committed state when no device fault is recorded for the target storage device and an error transition to the setup state when a device fault is recorded for the target storage device;

in the event of a clean system shutdown, completing a shutdown procedure in which a device fault detected during writing to the target storage device is recorded and the error flag is cleared, the recording of the device fault forcing the error transition of the migration operation during subsequent operation after system startup; and

during the system startup prior to enabling application input/output to the target storage device, determining whether the error flag is set indicating that the shutdown procedure was not completed, and only if the error flag is set then recording a device fault for the target storage device to force the error transition of the migration operation during subsequent operation.

2. A method according to claim 1, wherein the write cloning states include a source selected state in which application reads are directed to the source storage device, and wherein the device fault for the target storage device is recorded based on the migration operation being in the source selected state during the system startup.

3. A method according to claim 1, wherein;

the write cloning states include a target selected state in which application reads are directed to the target storage device;

the device fault for the target storage device is recorded, and the error transition is a first error transition made, based on the migration operation not being in the target selected state during the system startup; and

the allowed state transitions further include a second error transition to the committed state when a device fault is recorded for the source storage device;

and further including, during the system startup if the error flag is set and the migration operation is in the target selected state, recording a device fault for the source storage device to force the second error transition of the migration operation during subsequent operation.

4. A method according to claim 1, wherein the write cloning states include a source selected state in which application reads are directed to the source storage device and a target selected state in which application reads are directed to the target storage device, and the allowable transitions include transitions between the source selected state and the target selected state in the absence of device faults for the source and target storage devices, and wherein the device fault for the target storage device is recorded as a fault to the non-selected side when the migration operation is in the source selected state.

5. A method according to claim 1, further including, during the system startup after determining whether the error flag is set, setting the error flag for use in the event of another shutdown/startup cycle.

6. A method according to claim 1, performed in a computer in which application input/output is not enabled prior to commencement of user-space application execution, and wherein the startup procedure is a user-space procedure assuming no prior enablement of application input/output.

7. A method according to claim 1, performed in a computer in which application input/output is enabled prior to commencement of user-space application execution, and wherein determining whether the error flag is set and recording a device fault are performed by a kernel-space component of the startup procedure executed prior to enablement of application input/output.

8. A method according to claim 1, wherein the startup procedure includes, in the event that the error flag is set, automatically making the error transition of the migration operation prior to returning control of the migration operation to an administrative user.

9. A method according to claim 1, wherein the migration operation is one of a plurality of concurrently active migration operations, each active migration having its own respective error flag and device fault indicator, and wherein completing the shutdown procedure includes recording detected target device faults and clearing error flags of all the active migrations, and wherein the determining and recording performed during the system startup are performed for each of the active migrations.

10. A non-transitory computer readable medium having computer program instructions recorded thereon, the computer program instructions being executable by a computer to cause the computer to perform a method of managing state transitions in a data migration operation, the migration operation including an initial setup state, a set of write cloning states in which a target storage device is becoming or is maintained synchronized with a source storage device, and a later committed state in which the target storage device is used to the exclusion of the source storage device, the set of write cloning states spanning a system shutdown/startup cycle, the method including:

upon entering the write cloning states, setting an error flag used to protect against data corruption in the event of a non-clean system shutdown;

during the write cloning states, performing only allowed state transitions based on recorded device faults, the allowed state transitions including a non-error transition to the committed state when no device fault is recorded for the target storage device and an error transition to the setup state when a device fault is recorded for the target storage device;

in the event of a clean system shutdown, completing a shutdown procedure in which a device fault detected during writing to the target storage device is recorded and the error flag is cleared, the recording of the device fault forcing the error transition of the migration operation during subsequent operation; and

during the system startup prior to enabling application input/output to the target storage device, determining whether the error flag is set indicating that the shutdown procedure was not completed, and only if the error flag is set then recording a device fault for the target storage device to force the error transition of the migration operation during subsequent operation.

11. A non-transitory computer readable medium according to claim 10, wherein the write cloning states include a source selected state in which application reads are directed to the source storage device, and wherein the device fault for the target storage device is recorded based on the migration operation being in the source selected state during the system startup.

12. A non-transitory computer readable medium according to claim 10, wherein;

the write cloning states include a target selected state in which application reads are directed to the target storage device;

the device fault for the target storage device is recorded, and the error transition is a first error transition made, based on the migration operation not being in the target selected state during the system startup; and

the allowed state transitions further include a second error transition to the committed state when a device fault is recorded for the source storage device;

and wherein the method further includes, during the system startup if the error flag is set and the migration operation is in the target selected state, recording a device fault for the source storage device to force the second error transition of the migration operation during subsequent operation.

13. A non-transitory computer readable medium according to claim 10, wherein the write cloning states include a source selected state in which application reads are directed to the source storage device and a target selected state in which application reads are directed to the target storage device, and the allowable transitions include transitions between the source selected state and the target selected state in the absence of device faults for the source and target storage devices, and wherein the device fault for the target storage device is recorded as a fault to the non-selected side when the migration operation is in the source selected state.

14. A non-transitory computer readable medium according to claim 10, wherein the method further includes, during the system startup after determining whether the error flag is set, setting the error flag for use in the event of another shutdown/startup cycle.

15. A non-transitory computer readable medium according to claim 10, wherein the computer is a type in which application input/output is not enabled prior to commencement of user-space application execution, and wherein the startup procedure is a user-space procedure assuming no prior enablement of application input/output.

16. A non-transitory computer readable medium according to claim 10, wherein the computer is a type in which application input/output is enabled prior to commencement of user-space application execution, and wherein determining whether the error flag is set and recording a device fault are performed by a kernel-space component of the startup procedure executed prior to enablement of application input/output.

17. A non-transitory computer readable medium according to claim 10, wherein the startup procedure includes, in the event that the error flag is set, automatically making the error transition of the migration operation prior to returning control of the migration operation to an administrative user.

18. A non-transitory computer readable medium according to claim 10, wherein the migration operation is one of a plurality of concurrently active migration operations, each active migration having its own respective error flag and device fault indicator, and wherein completing the shutdown procedure includes recording detected target device faults and clearing error flags of all the active migrations, and wherein the determining and recording performed during the system startup are performed for each of the active migrations.