Sharing of Data Across Disjoint Clusters

• US 20110145243A1
• Filed: 12/16/2009
• Published: 06/16/2011
• Est. Priority Date: 12/16/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Yudenfriend, Harry M.
• Granted Patent: US 8,161,128 B2
• Time in Patent Office: Days
• US Class Current: 707/737
• CPC Class Codes: G06F 2209/523 Mode
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms

Abstract

Methods and devices are provided for sharing data across two or more different clusters. An operating system (OS) in a cluster checks a metadata record of a file system of a shared device to retrieve path group identifiers (PGIDs). A control unit list of the shared device is checked to retrieve PGIDs that are active on the shared device. An input/output supervisor (IOS) record in a couple dataset is checked to retrieve PGIDs in the cluster. The metadata record, control unit list, and IOS record are compared, and if a PGID is found in the metadata record that is not in the IOS record and if the found PGID is not in the control unit list, the found PGID is not active on the shared device. The found PGID of the different cluster is removed from metadata record, and members of the cluster can R/W to file system.

Claims

1. A method for sharing data across two or more different clusters, comprising:

checking, by an operating system in a cluster, a metadata record of a file system of a shared device to retrieve path group identifiers (PGIDs) in the metadata record of the file system, wherein the path group identifiers in the metadata record of the file system indicate that the path group identifiers are active in the file system, wherein the cluster comprises members and the operating system is one member of the cluster;

checking, by the operating system, a control unit list of the shared device to retrieve path group identifiers that are active on the shared device;

checking, by the operating system, an input/output supervisor (IOS) record in a couple dataset of the shared device to retrieve path group identifiers in the cluster, wherein the IOS record indicates that the path group identifiers are in the same cluster; and

comparing, by the operating system, the metadata record of the file system, the control unit list, and the IOS record in the couple dataset;

wherein based on the comparison if a PGID is found in the metadata record that is not in the IOS record by the operating system, the operating system determines that the found PGID is in a different cluster, and if the found PGID is not in the control unit list, the operating system determines that the found PGID is not active on the shared device;

wherein in response to the found PGID not being in the IOS record and not being in the control unit list, the operating system removes the found PGID of the different cluster from the metadata record of the file system; and

wherein in response to the found PGID of the different cluster being removed from the metadata record of the file system by the operating system, the operating system allows the members of the cluster to read and write to the file system when the found PGID is removed.

2. The method of claim 1, wherein if the found PGID in the metadata record of the file system is located in the IOS record by the operating system, the operating system determines that the found PGID is in the same cluster;

and

wherein the members of the cluster are operative to write to the file system.

3. The method of claim 1, wherein a customer specified policy for the file system can cause any shared access to data across different clusters to fail.

4. The method of claim 1, wherein the file system represents any data capable of being shared across separate clusters

5. A method for sharing data on a shared device across two or more different clusters, comprising:

storing, in an input/output supervisor (IOS) record of a shared device, path group identifiers (PGIDs) for members of a single cluster;

storing, in a control unit list of the shared device, path group identifiers that are active on the shared device; and

storing, in a metadata record of a file system of the shared device, path group identifiers that are actively accessing the file system;

wherein when the shared device causes the file system to enter a read multi-cluster state, the single cluster and different clusters can read the file system;

wherein when the shared device causes the file system to enter a read single cluster state, the single cluster can read the file system;

wherein when the shared device causes the file system to enter a read and write single cluster state, the single cluster can read and write to the file system; and

wherein if the shared device checks and has found a path group identifier (PGID) of the different clusters in the metadata record of the file system, if the shared device checks and does not locate the found PGID in the IOS record, and if the shared device checks and does not locate the found PGID in the control unit list, the shared device is operative to enter at least one of the read single cluster state and the read and write single cluster state.

6. The method of claim 5, wherein if the found PGID is in the metadata record of the file system and the found PGID is in the control list, the shared device is operative to enter the read multi-cluster state.

7. The method of claim 5, wherein the found PGID represents a plurality of PGIDs;

wherein the shared device is operative to check the metadata record of the file system, the IOS record, and the control list for each of the plurality of PGIDs;

wherein if the plurality of PGIDs are not located in the IOS record and if the plurality of PGIDs are not located in the control list, the shared device is operative to enter at least one of the read single cluster state and the read and write single cluster state; and

wherein the plurality of PGIDs are removed from the metadata record of the file system.

8. The method of claim 5, wherein the found PGID is removed from the metadata record of the file system on the shared device.

9. The method of claim 5, wherein when the shared device causes the file system to enter the read single cluster state, the different clusters can neither read nor write to the file system.

10. The method of claim 5, wherein when the shared device causes the file system to enter the read and write single cluster state, the different clusters neither read nor write to the file system.

11. A computing device coupled to a shared device for sharing data across two or more different clusters, comprising:

memory for storing a program; and

one or more processors, functionally coupled to the memory, the processor being responsive to computer-executable instructions contained in the program and operative for;

checking a metadata record of a file system of a shared device to retrieve path group identifiers (PGID) in the metadata record of the file system, wherein the path group identifiers in the metadata record of the file system indicate that the path group identifiers are active in the file system, wherein the cluster comprises members;

checking a control unit list of the shared device to retrieve path group identifiers that are active on the shared device;

checking an input/output supervisor (IOS) record in a couple dataset of the shared device to retrieve path group identifiers in the cluster, wherein the IOS record indicates that the path group identifiers are in the same cluster; and

comparing the metadata record of the file system, the control unit list, and the IOS record in the couple dataset;

wherein based on the comparison if a PGID is found in the metadata record that is not in the IOS record, the processor determines that the found PGID is in a different cluster, and if the found PGID is not in the control unit list, the processor determines that the found PGID is not active on the shared device;

wherein in response to the found PGID not being in the IOS record and not being in the control unit list, the processor removes the found PGID of the different cluster from the metadata record of the file system; and

wherein in response to the found PGID of the different cluster being removed from the metadata record of the file system, the processor allows the members of the cluster to read and write to the file system when the found PGID is removed.

12. The device of claim 11, wherein if the found PGID in the metadata record of the file system is located in the IOS record by the operating system, the processor determines that the found PGID is in the same cluster;

and

wherein the processor allows the members of the cluster to read and write to the file system.

13. A shared device for sharing data across two or more different clusters, comprising:

memory for storing a program; and

one or more processors, functionally coupled to the memory, the processor being responsive to computer-executable instructions contained in the program and operative for;

storing, in an input/output supervisor (IOS) record of a shared device, path group identifiers (PGIDs) for members of a single cluster;

storing, in a control unit list of the shared device, path group identifiers that are active on the shared device; and

storing, in a metadata record of a file system of the shared device, path group identifiers that are actively accessing the file system;

wherein when the processor causes the file system to enter a read multi-cluster state, the processor is operative to allow the single cluster and different clusters to read the file system;

wherein when the processor causes the file system to enter a read single cluster state, the processor is operative to allow the single cluster to read the file system;

wherein when the processor causes the file system to enter a read and write single cluster state, the processor is operative to allow the single cluster to read and write to the file system; and

wherein if the processor checks and has found a path group identifier (PGID) of the different clusters in the metadata record of the file system, if the processor checks and does not locate the found PGID in the IOS record, and if the processor checks and does not locate the found PGID in the control unit list, the processor is operative to enter at least one of the read single cluster state and the read and write single cluster state.

14. The device of claim 13, wherein if the found PGID is in the metadata record of the file system and the found PGID is in the control list, the processor is operative to enter the read multi-cluster state.

15. The device of claim 13, wherein the found PGID represents a plurality of PGIDs;

wherein the processor is operative to check the metadata record of the file system, the IOS record, and the control list for each of the plurality of PGIDs;

wherein if the plurality of PGIDs are not located in the IOS record and if the plurality of PGIDs are not located in the control list, the processor is operative to enter at least one of the read single cluster state and the read and write single cluster state; and

wherein processor removes the plurality of PGIDs from the metadata record of the file system.

16. The device of claim 13, wherein the processor removes the found PGID from the metadata record of the file system.

17. The device of claim 13, wherein when the processor causes the file system to enter the read single cluster state, the different clusters can neither read nor write to the file system.

18. The device of claim 13, wherein when the processor causes the file system to enter the read and write single cluster state, the different clusters neither read nor write to the file system.

19. A computer program product, tangibly embodied on a computer readable medium, the computer program product including instructions for causing a computer to execute a method for sharing data across two or more different clusters, comprising:

checking a metadata record of a file system of a shared device to retrieve path group identifiers (PGIDs) in the metadata record of the file system, wherein the path group identifiers in the metadata record of the file system indicate that the path group identifiers are active in the file system, wherein the cluster comprises members and the operating system is one member of the cluster;

checking a control unit list of the shared device to retrieve path group identifiers that are active on the shared device;

checking an input/output supervisor (IOS) record in a couple dataset of the shared device to retrieve path group identifiers in the cluster, wherein the IOS record indicates that the path group identifiers are in the same cluster; and

comparing the metadata record of the file system, the control unit list, and the IOS record in the couple dataset;

wherein based on the comparison if a PGID is found in the metadata record that is not in the IOS record, determine that the found PGID is in a different cluster, and if the found PGID is not in the control unit list, determine that the found PGID is not active on the shared device;

wherein in response to the found PGID not being in the IOS record and not being in the control unit list, remove the found PGID of the different cluster from the metadata record of the file system; and

wherein in response to the found PGID of the different cluster being removed from the metadata record of the file system, allow the members of the cluster to read and write to the file system when the found PGID is removed.

20. A computer program product, tangibly embodied on a computer readable medium, the computer program product including instructions for causing a computer to execute a method for sharing data on a shared device across two or more different clusters, comprising:

storing, in an input/output supervisor (IOS) record of a shared device, path group identifiers (PGIDs) for members of a single cluster;

storing, in a control unit list of the shared device, path group identifiers that are active on the shared device; and

storing, in a metadata record of a file system of the shared device, path group identifiers that are actively accessing the file system;

wherein when the shared device causes the file system to enter a read multi-cluster state, the single cluster and different clusters can read the file system;

wherein when the shared device causes the file system to enter a read single cluster state, the single cluster can read the file system;

wherein when the shared device causes the file system to enter a read and write single cluster state, the single cluster can read and write to the file system; and

wherein if the shared device checks and has found a path group identifier (PGID) of the different clusters in the metadata record of the file system, if the shared device checks and does not locate the found PGID in the IOS record, and if the shared device checks and does not locate the found PGID in the control unit list, the shared device is operative to enter at least one of the read single cluster state and the read and write single cluster state.