2 .\" Copyright 2006 by Cluster FileSystems. All Rights Reserved.
3 .\" This file may be copied under the terms of the GNU Public License.
5 .TH Lustre 7 "2006 Jun 15" Lustre "A high-performance cluster file system"
9 A high-performance file system designed for clusters.
12 is a high-performance, massively-scalable, POSIX-compliant network file system
13 designed for the world's largest high-performance compute clusters. Lustre
14 is under active development from Cluster FileSystems (http://clusterfs.com).
16 Lustre filesystems are made up of multiple servers typically distributed
17 across multiple nodes. Lustre clients can contact these server nodes over
18 multiple high-speed network fabrics via LNET, the Lustre NETworking
19 system. Clients then present the filesystem at a mount point.
21 A filesystem is comprised of a
23 , MetaData Target, which stores directory and file meta-information, and a
26 , Object Storage Targets, which hold the file data. There is typically not
27 a 1:1 mapping of a file on disk and what is presented as a file on a Lustre client.
29 Lustre and LNET are implemented as a series of kernel modules, for both
30 servers and clients. LNET networks are defined in the modprobe.conf file on
31 all nodes. Lustre and LNET make extensive use of /proc files for making
32 system data available, for example
35 to show the local LNET Network Interfaces.
37 .B /proc/fs/lustre/devices
38 to show locally running Lustre objects.
43 Format a physical disk for use as a Lustre server's backend storage (aka
47 Modify configuration information on a Lustre target disk.
52 that starts Lustre servers and clients.
55 A low-level interface to control various aspects of Lustre
58 A user-level interface to control Lustre-specific information for
61 Format multiple Lustre targets simultaneously from definitions in a CSV
64 Please report all bugs to Cluster FileSystems, support@clusterfs.com
68 filesystem package is available from CFS
73 .BR tunefs.lustre (8),