Global File System

from Wikipedia, the free encyclopedia

The Global File System ( GFS ) is a cluster file system that enables several computers to access shared memory at the same time and that ensures the consistency of the stored data. Often the shared storage is a storage area network (SAN). File locking ( English locking ) without a cluster file system does not work, takes a locking module of GFS.

GFS is part of the Linux kernel and is developed within its framework. The driving force is the Red Hat company .

history

GFS was originally developed as part of a feasibility study at the University of Minnesota . It was later taken over by Sistina Software, who developed it further as an open source project. In 2001 Sistina Software decided to further develop GFS as proprietary software . In 2003, Red Hat bought Sistina software and released GFS in 2004 with other cluster infrastructure programs under the GNU General Public License .

Technical

The GFS kernel module and a cluster configuration are required to mount GFS . In addition, further kernel modules and services are required for management (e.g. CMAN and ccsd), locking (e.g. lock_gulm and lock_dlm) and fencing (e.g. fenced) to exclude failed nodes.

The shared block storage devices are often made available over a network via a Global Network Block Device (GNBD) . The block devices can be imported to the cluster node with gnbd_import and then mounted.

GFS can use various locking mechanisms. The use of Distributed Lock Manager (DLM) is recommended . Red Hat Enterprise Linux prior to version 5 also included Grand Unified Lock Manager (GULM) , which is not recommended. The placeholder module Nolock can be used for local file systems that can be accessed exclusively .

With GFS as a basis, it is possible to set up a diskless shared root cluster .

Global network block device

Global Network Block Device (GNBD) is a service that provides a block device over the network. Devices exported in this way can be imported from several computers at the same time. These block devices, imported multiple times, often form the basis for GFS. Block devices that GNBD exports include hard disks , partitions and logical volumes of the Logical Volume Manager (LVM).

GNDB consists of kernel modules and utilities. It requires a functioning fencing (English for fencing , demarcating ). Fencing prevents I / O operations from nodes that, from the point of view of the cluster, are malfunctioning. Usually the fenced service is used for this.

Errors related to GFS

A common misconception is that cluster file systems can be used to export data over a network. The file system uses block storage devices (Engl. Block device ) to store data. In contrast to conventional file systems, cluster file systems do not assume exclusive use of the device. The way in which multiple systems access a block storage device at the same time is irrelevant to GFS. Possible techniques for sharing block devices include iSCSI and Serial Attached SCSI with multiple initiators.

See also

  • OCFS2 (Oracle Cluster File System 2)

Web links

Individual evidence

  1. ↑ FAQ / LockManager - Cluster Wiki. January 14, 2009, accessed November 14, 2010 .