QUICK INITIALIZATION OF DATA REGIONS IN A DISTRIBUTED STORAGE SYSTEM

• US 20160092468A1
• Filed: 09/30/2014
• Published: 03/31/2016
• Est. Priority Date: 09/30/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Hildebrand, Dean, Schmuck, Frank B.
• Granted Patent: US 9,916,325 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/1844 Management specifically adapted to replicated file systems
• CPC Class Codes: G06Q 10/00 Administration; Management

Abstract

A method includes flushing a write transaction based on byte-ranges within a file system block. It is determined if the write transaction is a first write transaction to the file system block and whether to flush the received write transaction to a recovery log. If it is determined to flush the write transaction to the recovery log and the received write transaction is the first write transaction: the data associated with the write transaction is appended in the recovery log and byte-ranges remaining in the file system block are recorded, and also recording an associated particular bit value or particular bit pattern that the remaining byte-ranges must be set to before the remaining byte-ranges are accessed or when a recovery sequence for recovering data in the recovery log executes due to failure. The write transaction is replicated. The entire file system block byte-range is marked as committed.

Claims

1. A method comprising:

flushing a received write transaction from a first memory device to a second memory device on a node in a file system based on one or more byte-ranges within a file system block;

determining if the received write transaction is a first write transaction for writing data associated with the received write transaction to the file system block;

determining to flush the received write transaction to a recovery log that is stored in a non-volatile storage layer of the file system;

if it is determined to flush the received write transaction to the recovery log and the received write transaction is the first write transaction for writing the data associated with the received write transaction to the file system block;

appending the data associated with the received write transaction in the recovery log and recording byte-ranges remaining in the file system block after writing the data associated with the write transaction to the file system block, and recording an associated particular bit value or a particular bit pattern that the remaining byte-ranges must be set to before the remaining byte-ranges are accessed or when a recovery sequence for recovering data in the recovery log executes due to a software or hardware failure;

replicating the data associated with the received write transaction as well as information regarding the remaining byte-ranges and the associated particular bit value or the particular bit pattern that must be set in another non-volatile storage layer of another node if required; and

marking the entire file system block byte-range as committed.

2. The method of claim 1, wherein determining to flush the received write transaction is based on one or more of:

an application has requested the flush, a timer has expired, to avoidance of data corruption or loss is required, and the data associated with the write transaction must be on the secondary memory device to maintain portable operating system interface (POSIX) semantics.

3. The method of claim 1, further comprising determining if a byte-range is to be placed in the recovery log based on whether the byte-range is not already marked as committed and a particular policy is met.

4. The method of claim 1, further comprising if the application updates a byte-range of a file system block in which at least a portion is marked as committed, clearing one or more committed markings on the byte-range of a new write transaction and marking the byte-range as dirty.

5. The method of claim 3, wherein the first memory device comprises a volatile memory device, and the second memory device comprises a primary storage device on the node of the file system.

6. The method of claim 5, further comprising:

upon node or file system failure, quiescing file system and flushing data requests from the non-volatile storage layer into the primary storage device and writing the associated particular bit value or the particular bit pattern to fill in the byte-ranges of the file system block on the primary storage device as recorded in the non-volatile storage layer, and then unquiescing the file system.

7. The method of claim 5, further comprising:

upon loss of access to a copy of data and recorded byte-ranges and the associated particular bit value or the particular bit pattern in the non-volatile storage layer that is on a different node than the node that is placing requests in the recovery log, re-replicating data to ensure correct number of copies.

8. The method of claim 5, further comprising:

integrating a file system recovery log layer in the file system, wherein the file system includes a distributed non-volatile storage layer that buffers data committed to a stable storage on cluster nodes of the file system;

flushing dirty and committed data to the primary storage device one of periodically or in response to a file system request, and marking the data no longer dirty; and

in response to memory or space pressure, or a timer expiring, traversing the recovery log and marking logged byte-ranges and byte-ranges with an associated particular bit value or particular bit pattern as no longer committed based on;

determining if logged byte-ranges are still dirty;

if the logged byte-ranges are still dirty, flushing data and writing the associated particular bit value or the particular bit pattern to fill in the byte-ranges of the file system block on the primary storage device as recorded in the distributed non-volatile storage layer to the primary storage device; and

marking the logged byte-ranges as no longer dirty; and

removing byte-range records and bit value or bit pattern records from the recovery log.

9. The method of claim 8, further comprising:

in response to a read transaction for dirty data located on a file system node other than a node on which the dirty data is present, coordinating flushing the dirty data to a primary storage device on the node with the dirty data prior to satisfying the read transaction.

10. A computer program product for quick initialization using storage and replication, the computer program product comprising a computer readable storage medium having program code embodied therewith, the program code readable/executable by a processor to:

flush a received write transaction from a first memory device to a second memory device on a node in a file system based on one or more byte-ranges within a file system block;

determine if the received write transaction is a first write transaction for writing data associated with the received write transaction to the file system block;

determine to flush the received write transaction to a recovery log that is stored in a non-volatile storage layer of the file system;

if it is determined to flush the received write transaction to the recovery log and the received write transaction is the first write transaction for writing the data associated with the received write transaction to the file system block;

append the data associated with the received write transaction in the recovery log and record byte-ranges remaining in the file system block after writing the data associated with the write transaction to the file system block, and recording an associated particular bit value or a particular bit pattern that the remaining byte-ranges must be set to before the remaining byte-ranges are accessed or when a recovery sequence for recovering data in the recovery log executes due to a software or hardware failure;

replicate the data associated with the write transaction as well as information regarding the remaining byte-ranges and the associated particular bit value or the particular bit pattern that must be set in another non-volatile storage layer of another node if required; and

mark the entire file system block byte-range as committed.

11. The computer program product of claim 10, wherein determining to flush the received write transaction is based on whether an application has requested the flush, a timer has expired, avoidance or data corruption or loss is required, or the data associated with the write transaction must be on the secondary memory device to maintain portable operating system interface (POSIX) semantics.

12. The computer program product of claim 10, further comprising:

determine if a byte-range is to be placed in the recovery log based on whether the byte-range is not already marked as committed and a particular policy is met.

13. The computer program product of claim 10, further comprising if the application updates a byte-range of a file system block in which at least a portion is marked as committed, clearing one or more committed markings on the byte-range of a new write transaction and marking the byte-range as dirty, and the first memory device comprises a volatile memory device, and the second memory device comprises a primary storage device on the node of the file system.

14. The computer program product of claim 14, further comprising:

upon node or file system failure, quiesce file system and flushing data requests from the non-volatile storage layer into the primary storage device and write the associated particular bit value or the particular bit pattern to fill in the byte-ranges of the file system block on the primary storage device as recorded in the non-volatile storage layer, and then unquiesce the file system.

15. The computer program product of claim 14, further comprising:

upon loss of access to a copy of the data associated with the write transaction and the recorded byte-ranges remaining in the file system block after writing the data associated with the write transaction to the file system block and the associated particular bit value or the particular bit pattern in the non-volatile storage layer that is on a different node than the node that is placing requests in the recovery log, re-replicate the data associated with the write transaction to ensure correct number of copies;

integrating a file system recovery log layer in the file system, wherein the file system includes a distributed non-volatile storage layer that buffers data committed to a stable storage on cluster nodes of the file system;

flushing dirty and committed data to the primary storage device one of periodically or in response to a file system request, and marking the dirty data no longer dirty; and

in response to memory or space pressure, or a timer expiring, traversing the recovery log and marking logged byte-ranges and byte-ranges with the associated particular bit value or the particular bit pattern as no longer committed based on;

determining if logged byte-ranges are still dirty;

if the logged byte-ranges are still dirty, flushing the dirty data and writing the associated particular bit value or the particular bit pattern to fill in the byte-ranges of the file system block on the primary storage device as recorded in the distributed non-volatile storage layer to the primary storage device; and

marking the logged byte-ranges as no longer dirty; and

removing byte-range records and bit value or bit pattern records from the recovery log.

16. The computer program product of claim 15, further comprising:

in response to a read transaction for dirty data located on a file system node other than a node on which the dirty data is present, coordinate flushing the dirty data to a primary storage device on the node with the dirty data prior to satisfying the read transaction.

17. A system comprising:

a file system including a primary storage device; and

a client node device coupled with a recovery log and a first memory device, the client node including a process that;

flushes a received write transaction from the first memory device to the primary memory device on a node in the file system based on one or more byte-ranges for a file system block, determines if the received write transaction is a first write transaction for writing data associated with the received write transaction to the file system block, determines to flush the received write transaction to a recovery log that is stored in a non-volatile storage layer of the file system, and if it is determined to flush the received write transaction from the recovery log;

appends the data associated with the received write transaction in the recovery log and records byte-ranges remaining in the file system block after writing the data associated with the write transaction to the file system block, and records an associated particular bit value or particular bit pattern that the remaining byte-ranges must be set to before the remaining byte-ranges are accessed or when a recovery sequence for recovering data in the recovery log executes due to a software or hardware failure, replicates the data associated with the write transaction as well as information regarding the remaining byte-ranges and the associated particular bit value or the particular bit pattern that must be set in another non-volatile storage layer of another node if required, and marks the entire file system block byte-range as committed.

18. The system of claim 17, further comprising if the process updates a byte-range of a file system block in which at least a portion is marked as committed, the process clears one or more committed markings on the byte-range of a new write transaction and marks the byte-range as dirty, wherein the system further comprising:

upon node or file system failure, the process quiesces file system and flushing data requests from the non-volatile storage layer into the primary storage device and writing the associated particular bit value or the particular bit pattern to fill in the byte-ranges of the file system block on the primary storage device as recorded in the non-volatile storage layer, and then unquiesces the file system.

19. The system of claim 18, further comprising:

upon loss of access to a copy of data and recorded byte-ranges and the associated particular bit value or the particular bit pattern in the non-volatile storage layer that is on a different node than the node that is placing requests in the recovery log, the process re-replicates data to ensure correct number of copies.

20. The system of claim 18, wherein the process further:

integrates a file system recovery log layer in the file system wherein the file system includes a distributed non-volatile storage layer that buffers data committed to a stable storage on cluster nodes of the file system;

flushes dirty and committed data to the primary storage device one of periodically or in response to a file system request, and marking the dirty data no longer dirty; and

in response to memory or space pressure, or a timer expiring, traverses the recovery log and marks logged byte-ranges and byte-ranges with the associated particular bit value or the particular bit pattern as no longer committed based on the process;

determining if logged byte-ranges are still dirty;

if the logged byte-ranges are still dirty, flushing the dirty data and writing the associated particular bit value or the particular bit pattern to fill in the byte-ranges of the file system block on the primary storage device as recorded in the distributed non-volatile storage layer to the primary storage device; and

marking the logged byte-ranges as no longer dirty; and

removing byte-range records and bit value or bit pattern records from the recovery log.