Processing incomplete data access transactions

• US 10,157,021 B2
• Filed: 06/29/2016
• Issued: 12/18/2018
• Est. Priority Date: 06/29/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Baptist, Andrew D., Dhuse, Greg R., Khadiwala, Ravi V.
• Primary Examiner(s): Chappell, Daniel C
• Application Number:
  US15/196,442
• Publication Number:
  US 20180004453A1
• Time in Patent Office: 902 Days
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]

Abstract

A method includes obtaining a set of pending transaction information from a set of storage units regarding a plurality of pending transactions. A pending transaction information includes information regarding an encoded data slice and a computing device issuing a data access request. The method further includes identifying an incomplete transaction based on the set of pending transaction information. The method further includes determining whether to complete the incomplete transaction based on information regarding the requesting computing device. The method further includes, when the incomplete transaction is not to be completed, instructing the storage units to discard a corresponding data access request associated with the incomplete transaction. The method further includes, when the incomplete transaction is to be completed, determining what aspects of the corresponding data access request have not been completed and instructing at least one storage unit to complete the aspects of the corresponding data access request.

Claims

1. A method for execution by a computing device of a dispersed storage network (DSN), the method comprises:

obtaining a set of pending transaction information from a set of storage units of the DSN regarding a plurality of pending transactions, wherein a pending transaction of the plurality of pending transactions is regarding a data access request to a set of encoded data slices and wherein the pending transaction information from a storage unit of the set of storage units is regarding one or more pending data access requests for an encoded data slice of the set of encoded data slices, wherein a data segment of a data object is dispersed storage error encoded to produce the set of encoded data slices, and wherein the set of encoded data slices are stored in the set of storage units;

identifying an incomplete transaction based on the set of pending transaction information, wherein the incomplete transaction is a stalled pending transaction of the plurality of pending transactions;

determining whether to complete the incomplete transaction based on information regarding a requesting computing device associated with the incomplete transaction;

when the incomplete transaction is not to be completed, instructing the storage units to discard a corresponding data access request associated with the incomplete transaction; and

when the incomplete transaction is to be completed;

determining what aspects of the corresponding data access request have not been completed; and

instructing at least one of the storage units to complete the aspects of the corresponding data access request.

2. The method of claim 1, wherein the pending transaction information comprises one or more of:

one or more proposal records, wherein a proposal record of the one or more proposal records includes, for an encoded data slice or a batch of encoded data slices;

an ordered list of pending transactions regarding the encoded data slice or the batch of encoded data slices;

an indication of current revision level for the encoded data slice or the batch of encoded data slices; and

an ordered list of different version of the encoded data slice or the batch of encoded data slices;

a status table that includes information regarding status of data access requesting computing devices having one or more pending transactions; and

a blacklist that includes identity of inactive computing devices, wherein an inactive computing device of the inactive computing devices is affiliated with a transaction that cannot be completed due to a persistent inactive status of the inactive computing device.

3. The method of claim 1 further comprises:

identifying the incomplete transaction is a write operation regarding the set of encoded data slices from the requesting computing device, wherein a write threshold number of encoded data slices of the set of encoded data slices have been successfully stored in at least some of the storage units of the set of storage unit, wherein a particular encoded data slice of the set of encoded data slices was not successfully finalized by a particular storage unit of the set of storage units, and wherein the unsuccessful finalizing of the particular encoded data slice corresponds to the aspects of the corresponding data access request that have not been completed;

determining whether the particular storage unit was sent a finalize request regarding the particular encoded data slice by the requesting computing device;

when the particular storage unit was not sent the finalize request, resending the finalize request to the particular storage unit to complete the write operation;

when the particular storage unit was sent the finalize request;

determining whether the particular storage unit is currently available to receive the finalize request;

when the particular storage unit is available to receive the finalize request, resending the finalize request to the particular storage unit to complete the write operation; and

when the particular storage unit is not available to receive the finalize request, flagging the particular encoded data slice for rebuilding.

4. The method of claim 3 further comprises:

indicating that the incomplete transaction is not to be completed by the particular storage unit when the particular encoded data slice is flagged for rebuilding.

5. The method of claim 1 further comprises:

identifying the incomplete transaction is a write operation regarding the set of encoded data slices from the requesting computing device, wherein the requesting computing device has a threshold number of write requests of a set of write requests regarding the set of encoded data slices in a first priority position in a set of proposal records of the pending transaction information from the set of storage units, wherein the requesting computing device has not issued a set of finalize requests to the set of storage units regarding the set of encoded data slices, and wherein the not issuing the set of finalize requests corresponds to the aspects of the corresponding data access request that have not been completed;

determining status of the requesting computing device based on the pending transaction information; and

when a majority of the pending transaction information indicates that the requesting computing device has a status of inactive, indicating that the incomplete transaction is not to be completed.

6. The method of claim 5 further comprises:

when a majority of the pending transaction information indicates that the requesting computing device has a status of active, determining whether to issue the set of finalize requests on behalf of the requesting computing device; and

when the set of finalize requests are to be sent on behalf of the requesting computing device, sending the set of finalize requests to the set of storage units.

7. The method of claim 6 further comprises:

when the set of finalize requests are not to be sent on behalf of the requesting computing device, sending an indication to the set of storage units to keep the set of write requests pending.

8. The method of claim 1 further comprises:

identifying the incomplete transaction is a write operation regarding the set of encoded data slices from the requesting computing device, wherein the requesting computing device does not have a threshold number of write requests of a set of write requests regarding the set of encoded data slices in a first priority position in a set of proposal records of the pending transaction information from the set of storage units, wherein the requesting computing device has not issued a set of undo requests to the set of storage units regarding the set of encoded data slices, and wherein the not issuing the set of undo requests and the set of write requests that are pending correspond to the aspects of the corresponding data access request that have not been completed;

determining status of the requesting computing device based on the pending transaction information; and

when a majority of the pending transaction information indicates that the requesting computing device has a status of inactive, indicating that the incomplete transaction is not to be completed and sending a set of undo requests to the set of storage units.

9. The method of claim 1 further comprises:

identifying the incomplete transaction as a write operation regarding the set of encoded data slices from the requesting computing device, wherein the requesting computing device is involved in a process of establishing one set of a plurality of sets of write requests as a winning set of write requests having a threshold number of write requests in a first priority position and wherein the process of establishing the winning set corresponds to the aspects of the corresponding data access request that have not been completed;

determining status of the requesting computing device based on the pending transaction information;

when a majority of the pending transaction information indicates that the requesting computing device has a status of inactive, indicating that the incomplete transaction is not to be completed and sending a set of undo requests to the set of storage units; and

when the majority of the pending transaction information indicates that the requesting computing device has the status of active;

indicating that the incomplete transaction is to be completed; and

instructing the set of storage units to continue with the processing of establishing the winning set.

10. The method of claim 1 further comprises:

after completion of the incomplete transaction, sending a set of clean up requests to the set of storage units.

11. A computer readable memory comprises:

a first memory element that stores operational instructions that, when executed by a computing device, causes the computing device to;

obtain a set of pending transaction information from a set of storage units of a dispersed storage network (DSN) regarding a plurality of pending transactions, wherein a pending transaction of the plurality of pending transactions is regarding a data access request to a set of encoded data slices and wherein the pending transaction information from a storage unit of the set of storage units is regarding one or more pending data access requests for an encoded data slice of the set of encoded data slices, wherein a data segment of a data object is dispersed storage error encoded to produce the set of encoded data slices, and wherein the set of encoded data slices are stored in the set of storage units;

a second memory element that stores operational instructions that, when executed by the computing device, causes the computing device to;

identify an incomplete transaction based on the set of pending transaction information, wherein the incomplete transaction is a stalled pending transaction of the plurality of pending transactions; and

determine whether to complete the incomplete transaction based on information regarding a requesting computing device associated with the incomplete transaction;

a third memory element that stores operational instructions that, when executed by the computing device, causes the computing device to;

when the incomplete transaction is not to be completed, instruct the storage units to discard a corresponding data access request associated with the incomplete transaction; and

when the incomplete transaction is to be completed;

determine what aspects of the corresponding data access request have not been completed; and

instruct at least one of the storage units to complete the aspects of the corresponding data access request.

12. The computer readable memory of claim 11, wherein the pending transaction information comprises one or more of:

one or more proposal records, wherein a proposal record of the one or more proposal records includes, for an encoded data slice or a batch of encoded data slices;

an ordered list of pending transactions regarding the encoded data slice or the batch of encoded data slices;

an indication of current revision level for the encoded data slice or the batch of encoded data slices; and

an ordered list of different version of the encoded data slice or the batch of encoded data slices;

a status table that includes information regarding status of data access requesting computing devices having one or more pending transactions; and

a blacklist that includes identity of inactive computing devices, wherein an inactive computing device of the inactive computing devices is affiliated with a transaction that cannot be completed due to a persistent inactive status of the inactive computing device.

13. The computer readable memory of claim 11 further comprises:

the second memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

identify the incomplete transaction as a write operation regarding the set of encoded data slices from the requesting computing device, wherein a write threshold number of encoded data slices of the set of encoded data slices have been successfully stored in at least some of the storage units of the set of storage unit, wherein a particular encoded data slice of the set of encoded data slices was not successfully finalized by a particular storage unit of the set of storage units, and wherein the unsuccessful finalizing of the particular encoded data slice corresponds to the aspects of the corresponding data access request that have not been completed; and

determine whether the particular storage unit was sent a finalize request regarding the particular encoded data slice by the requesting computing device;

the third memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

when the particular storage unit was not sent the finalize request, resending the finalize request to the particular storage unit to complete the write operation; and

when the particular storage unit was sent the finalize request;

determining whether the particular storage unit is currently available to receive the finalize request;

when the particular storage unit is available to receive the finalize request, resending the finalize request to the particular storage unit to complete the write operation; and

when the particular storage unit is not available to receive the finalize request, flagging the particular encoded data slice for rebuilding.

14. The computer readable memory of claim 13, wherein the third memory element further stores operational instructions that, when executed by the computing device, causes the computing device to:

indicate that the incomplete transaction is not be completed by the particular storage unit when the particular encoded data slice is flagged for rebuilding.

15. The computer readable memory of claim 11 further comprises:

the second memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

identify the incomplete transaction is a write operation regarding the set of encoded data slices from the requesting computing device, wherein the requesting computing device has a threshold number of write requests of a set of write requests regarding the set of encoded data slices in a first priority position in a set of proposal records of the pending transaction information from the set of storage units, wherein the requesting computing device has not issued a set of finalize requests to the set of storage units regarding the set of encoded data slices, and wherein the not issuing the set of finalize requests corresponds to the aspects of the corresponding data access request that have not been completed; and

determine status of the requesting computing device based on the pending transaction information; and

the third memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

when a majority of the pending transaction information indicates that the requesting computing device has a status of inactive, indicate that the incomplete transaction is not to be completed.

16. The computer readable memory of claim 15, wherein the third memory element further stores operational instructions that, when executed by the computing device, causes the computing device to:

when a majority of the pending transaction information indicates that the requesting computing device has a status of active, determine whether to issue the set of finalize requests on behalf of the requesting computing device; and

when the set of finalize requests are to be sent on behalf of the requesting computing device, send the set of finalize requests to the set of storage units.

17. The computer readable memory of claim 16, wherein the third memory element further stores operational instructions that, when executed by the computing device, causes the computing device to:

when the set of finalize requests are not to be sent on behalf of the requesting computing device, send an indication to the set of storage units to keep the set of write requests pending.

18. The computer readable memory of claim 11 further comprises:

the second memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

identify the incomplete transaction is a write operation regarding the set of encoded data slices from the requesting computing device, wherein the requesting computing device does not have a threshold number of write requests of a set of write requests regarding the set of encoded data slices in a first priority position in a set of proposal records of the pending transaction information from the set of storage units, wherein the requesting computing device has not issued a set of undo requests to the set of storage units regarding the set of encoded data slices, and wherein the not issuing the set of undo requests and the set of write requests that are pending correspond to the aspects of the corresponding data access request that have not been completed; and

determine status of the requesting computing device based on the pending transaction information;

the third memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

when a majority of the pending transaction information indicates that the requesting computing device has a status of inactive, indicate that the incomplete transaction is not to be completed and sending a set of undo requests to the set of storage units.

19. The computer readable memory of claim 11 further comprises:

the second memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

identify the incomplete transaction is a write operation regarding the set of encoded data slices from the requesting computing device, wherein the requesting computing device is involved in a process of establishing one set of a plurality of sets of write requests as a winning set of write requests having a threshold number of write requests in a first priority position and wherein the process of establishing the winning set corresponds to the aspects of the corresponding data access request that have not been completed; and

determine status of the requesting computing device based on the pending transaction information;

the third memory element further storing operational instructions that, when executed by the computing device, causes the computing device to;

when a majority of the pending transaction information indicates that the requesting computing device has a status of inactive, indicate that the incomplete transaction is not to be completed and sending a set of undo requests to the set of storage units; and

when the majority of the pending transaction information indicates that the requesting computing device has the status of active;

indicate that the incomplete transaction is to be completed; and

instruct the set of storage units to continue with the processing of establishing the winning set.

20. The computer readable memory of claim 11, wherein the third memory element further stores operational instructions that, when executed by the computing device, causes the computing device to:

after completion of the incomplete transaction, send a set of clean up requests to the set of storage units.