In-flight block map for a clustered redirect-on-write filesystem

• US 8,904,006 B2
• Filed: 12/08/2010
• Issued: 12/02/2014
• Est. Priority Date: 12/08/2010
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chang, Joon, Palsule, Ninad S., Solomon, Andrew N.
• Primary Examiner(s): NGUYEN, MINH CHAU
• Application Number:
  US12/963,078
• Publication Number:
  US 20120151245A1
• Time in Patent Office: 1,455 Days
• Field of Search: 709201-203, 709224-226
• US Class Current: 709/226
• CPC Class Codes: G06F 16/128 Details of file system snapshots on the file-level, e.g. snapshot creation, administration, deletion use of snapshots for error detection or correction G06F11/14, G06F11/16
• CPC Class Codes: G06F 16/1865 Transactional file systems

Abstract

A cluster server manages allocation of free blocks to cluster clients performing writes in a clustered file system. The cluster server manages free block allocation with a free block map and an in-flight block map. The free block map is a data structure or hardware structure with data that indicates blocks or extents of the clustered file system that can be allocated to a client for the client to write data. The in-flight block map is a data structure or hardware structure with data that indicates blocks that have been allocated to clients, but remain in-flight. A block remains in-flight until the clustered file system metadata has been updated to reflect a write performed to that block by a client. After a consistency snapshot of the metadata is published to the storage resources, the data at the block will be visible to other nodes of the cluster.

Claims

1. A computer program product for allocating cluster storage units to cluster nodes for a clustered file system that implements redirect-on-write snapshotting, the computer program product comprising:

a computer readable storage medium having program instructions embodied therewith, the program instructions comprising program instructions to,

track cluster storage units for a clustered file system of a cluster that implements redirect-on-write snapshotting as free, in-flight, or committed, wherein the cluster storage units comprise logical units of aggregate storage space of storage devices of the cluster;

indicate a first set of the cluster storage units as free because the first set of cluster storage units are available for allocation to nodes of the cluster;

receive a request for cluster storage units from a client node;

in response to reception of the request for cluster storage units from the client node,

transition the first set of the cluster storage units from free to in-flight to represent that the first set of cluster storage units are allocated to the client node for the client node to write to the first set of the cluster storage units; and

send first indications of the first set of the cluster storage units to the client node,

receive, from the client node, second indications of the first set of the cluster storage units, checksums corresponding to the first set of the cluster storage units, and an indication of a fileset, wherein reception of the second indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage unit indicates that the client node has written data to the first set of the cluster storage units;

in response to reception of, from the client node, the second indications of the first set of the cluster storage units and checksums corresponding to the first set of the cluster storage units, write the second indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage units to a location in metadata for the clustered file system corresponding to the fileset;

if a failure does not interrupt a writing of the indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage units to a location in metadata for the clustered file system corresponding to the fileset, then transition the first set of the cluster storage units to committed; and

if a failure interrupts a writing of the indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage units to a location in metadata for the clustered file system corresponding to the fileset, then return the first set of the cluster storage units to free.

2. The computer program product of claim 1, wherein said program instructions to indicate the first set of the cluster storage units as free comprises program instructions to represent the first set of the cluster storage units in a first structure that comprises information that identifies those of the cluster storage units that are available for allocation to nodes.

3. The computer program product of claim 2, wherein said program instructions to transition the first set of the cluster storage units from free to in-flight comprises program instructions to:

remove information that identifies the first set of the cluster storage units from the first structure; and

add the information that identifies the first set of the cluster storage units to a second structure that comprises information that identifies those of the cluster storage units that have been allocated to nodes for writing.

4. The computer program product of claim 3, wherein said program instructions to transition the first set of the cluster storage units to committed comprises program instructions to:

remove the information that identifies the first set of the cluster storage units from the second structure; and

record the checksums into the metadata for the clustered file system, wherein the client node generated and communicated the checksums.

5. The computer program product of claim 1, wherein the program instructions further comprise program instructions to:

detect a failure at the client node before at least one of the client node writes the data to the cluster storage units, the data written to the cluster storage units by the client node are visible to the cluster, and updating of the metadata with the checksums completes,

wherein said program instructions to return the first set of the cluster storage units to free comprises program instructions to;

remove information that identifies the first set of the cluster storage units from a first structure that comprises information that identifies those of the cluster storage units that have been allocated; and

update a second structure with the information that identifies the first set of the cluster storage units, wherein the second structure comprises information that identifies those of the cluster storage units that are available for allocation.

6. The computer program product of claim 1, wherein the program instructions further comprise program instructions to:

detect that a first server node that managed a first fileset of the clustered file system failed, wherein the computer is delegated management responsibility for the first fileset in response to the failure of the first server node;

load free cluster storage unit data stored to the aggregate storage space by the first server node, wherein the free cluster storage unit data indicates those of the cluster storage units that were free before the first server node failed;

load in-flight cluster storage unit data stored to the aggregate storage space by the first server node, wherein the in-flight cluster storage unit data indicates those of the cluster storage units that were in-flight before the first server node failed;

return those of the cluster storage units indicated as in-flight by the in-flight cluster storage unit data to free;

query client nodes of the cluster to determine those of the cluster storage units considered in-flight by the client nodes; and

transition those of the cluster storage units considered as in-flight by the client nodes from free to in-flight based on responses from the client nodes.

7. A computer program product for allocating cluster storage units to cluster nodes for a clustered file system that implements redirect-on-write snapshotting, the computer program product comprising:

a computer readable storage medium having program instructions embodied therewith, the program instructions comprising program instructions to,

receive a request for a number of cluster storage units from a client node in a cluster that implements redirect-on-write snapshotting for a clustered file system;

in response to reception of the request for the number of cluster storage units from the client node in the cluster that implements redirect-on-write snapshotting for a clustered file system, determine a set of cluster storage units that are free and that satisfy the number of cluster storage units;

indicate the set of cluster storage units as allocated to the client node and no longer free;

allocate the set of cluster storage units to the client node;

send first indications of the set of the cluster storage units to the client node;

receive a plurality of checksums and second indications of the set of cluster storage units from the client node, wherein the client node writes data to the set of cluster storage units and generates the plurality of checksums for the data; and

in response to reception of the plurality of checksums and the second indications of the set of cluster storage units from the client node,

update metadata for the clustered file system to indicate the plurality of checksums for the data written to the set of cluster storage units; and

update the metadata for the clustered file system to indicate an association between a fileset and the second indications of the set of cluster storage units.

8. The computer program product of claim 7, wherein the program instructions further comprise program instructions to remove an indication that the set of cluster storage units are allocated if an operation to update the metadata for the clustered file system completes successfully.

9. The computer program product of claim 8, wherein said program instructions to remove an indication that the set of cluster storage units are allocated comprises program instructions to remove data that identifies the set of cluster storage units from a structure that indicates cluster storage units allocated to nodes for the nodes to write.

10. The computer program product of claim 7, wherein said program instructions to indicate the set of cluster storage units as allocated to the client node and no longer free comprises program instructions to:

remove data that identifies the set of cluster storage units from a first structure that indicates cluster storage units available for allocation to nodes of the cluster; and

insert the data into a second structure that indicates cluster storage units allocated to nodes.

11. An apparatus comprising:

a processor;

a network interface configured to transmit and receive data to nodes of a cluster; and

a computer readable storage medium coupled with the processor, the computer readable storage medium having computer usable program code embodied therewith, the computer usable program code, which, when executed by the processor, cause the apparatus to,

track cluster storage blocks for a clustered file system of the cluster that implements redirect-on-write snapshotting as free, in-flight, or committed, wherein the cluster storage blocks comprise logical units of aggregate storage space of storage devices of the cluster;

indicate a first set of the cluster storage blocks as free because the first set of cluster storage blocks are available for allocation to nodes of the cluster;

receive a request for cluster storage units from a client node;

in response to reception of the request for cluster storage units from the client node, transition the first set of the cluster storage blocks from free to in-flight to represent that the first set of cluster storage blocks are allocated to the client node for the client node to write to the first set of the cluster storage blocks; and

send first indications of the first set of the cluster storage units to the client node;

receive, from the client node, second indications of the first set of the cluster storage units, checksums corresponding to the first set of the cluster storage units, and an indication of a fileset, wherein receiving the second indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage unit indicates that the client node has written data to the first set of the cluster storage units;

in response to reception, from the client node, of the second indications of the first set of the cluster storage units and checksums corresponding to the first set of the cluster storage units, write the second indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage units to a location in metadata for the clustered file system corresponding to the fileset;

if a failure does not interrupt a writing of the indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage units to a location in metadata for the clustered file system corresponding to the fileset, then transition the first set of the cluster storage blocks to committed; and

if a failure interrupts a writing of the indications of the first set of the cluster storage units and the checksums corresponding to the first set of the cluster storage units to a location in metadata for the clustered file system corresponding to the fileset, then return the first set of the cluster storage blocks to free.

12. The apparatus of claim 11, wherein the computer usable program code which, when executed by the processor, further causes the apparatus to:

detect that a first server node that managed a first fileset of the clustered file system failed, wherein the apparatus is delegated management responsibility for the first fileset in response to the failure of the first server node;

load free cluster storage unit data stored to the aggregate storage space by the first server node, wherein the free cluster storage unit data indicates those of the cluster storage units that were free before the first server node failed;

load in-flight cluster storage unit data stored to the aggregate storage space by the first server node, wherein the in-flight cluster storage unit data indicates those of the cluster storage units that were in-flight before the first server node failed;

return those of the cluster storage units indicated as in-flight by the in-flight cluster storage unit data to free;

query client nodes of the cluster to determine those of the cluster storage units considered in-flight by the client nodes; and

transition those of the cluster storage units considered as in-flight by the client nodes from free to in-flight based on responses from the client nodes.