System And Method For Hierarchical Recovery Of A Cluster File System
AssigneeInternational Business Machines Corporation
InventorsRobert K. Gjertsen, Manoj N. Kumar, David J. Craft, Gokul B. Kandiraju
Life of this patent
Estimating the term…
Abstract
Hierarchical recovery of failed computing nodes to operative computing nodes within a cluster of computing nodes is managed by initiating a recovery leader at an operative node that retrieves management information from the operative nodes and applies the management information to recover filesets of a meta-fileset in a hierarchical filesystem. The use of hierarchical filesets throughout the cluster provides more rapid failover by distributing recovery load across computing nodes and supporting recovery of nodes in parallel.
Claims20
A method for recovery of one or more computing nodes of a cluster having plural computing nodes managed with a meta-fileset and plural filesets, the method comprising:
- detecting failure of one or more computing nodes;
- designating one of the computing nodes as a recovery leader;
- retrieving to the recovery leader management information stored at the computing nodes;
- selecting with the recovery node one of the computing nodes as a file system manager;
- establishing the meta-fileset at the file system manager;
- selecting one of the computing nodes as a fileset manager for each of the plural filesets;
- establishing each of the filesets at its selected computing node;
- and
- updating the computing nodes to point to the file system manager and fileset managers.
A method for recovery of one or more failed computing nodes in a cluster having hierarchical file management, the method comprising:
- selecting a computing node to act as recovery leader of the cluster;
- retrieving management information to the recovery leader from each computing node, the management information including at least the filesets accessed by each computing node and the filesets served by each computing node;
- and
- applying the management information to simultaneously recover the one or more failed computing nodes to plural operative computing nodes.
A system for recovering one or more failed computing nodes of a cluster having plural computing nodes to an operative computing node of the cluster, the system comprising:
- a recovery module stored on each of plural of the computing nodes and operable to select a computing node to execute a recovery leader;
- and
- a recovery leader executing on the selected computing node and operable to retrieve management information from operative computing node of the cluster to select a computing node to execute a filesystem manager associated with a meta-filesystem and to execute one or more fileset managers, each fileset manager associated with a fileset.
Dependencies are inferred from the claim text.