| PRIMECLUSTER Global File Services Configuration and Administration Guide 4.2 (Solaris(TM) Operating Environment) |
|
Contents
Index
|
This interface can handle files that exceed 2 gigabytes.
For details, see largefile(5) in the "Solaris X Reference Manual Collection" or the online manual.
Access client. Module which operates in the kernel layer (file system layer) on each node sharing GFS Shared File System. An AC communicates with the MDS if a user application operating on the node sends an access request in a system call. File operations are performed using meta-data sent from the MDS and direct access to the real.
ACL is a file security system that specifies file access authority for specific users or groups in addition to the owner and owner group. For details, see setfacl(1) in the "Solaris X Reference Manual Collection" or the online manual.
The file system that made on the shared disk device is exclusive used in active node and the standing node in RMS Cluster.
Area management data includes information on i-nodes, V-data, and file data allocation information. It is stored in the meta-data area.
Disk file unit in which two or more physical disks are grouped as a single virtual disk. By dividing access between two or more physical disks, performance improves and high reliability can be implemented. Ordinarily, two or more access methods, which are called redundant array of inexpensive disks (RAID) levels, are provided. The major RAID levels are RAID 0 (striping), RAID 1 (mirroring), RAID 5 (striping with parity), and RAID 0+1 (striping set mirroring).
If a block failure on a disk device destroys partition structure information, a copy is available for continued processing of the file system.
A cylinder group consists of a single disk cylinder or multiple contiguous disk cylinders that have i-node slots for files.
The term data refers to file data.
The term data area refers to a file data area.
The term data area refers to a file data partition.
Direct I/O refers to a mode of data input/output that does not use kernel buffering.
The abbreviation for Equivalent Copy. The hardware function of the disk array equipment (ETERNUS6000 and ETERNUS3000). This function is also found in EMC's storage device Symmetrix. EC is the function to hold the mirror state between disks within hardware. The duplicate of a disk is created by separation operation of a mirror state. In the case of EC employment in high-speed backup restoration of the GFS Shared File System by sfcproxyop(1M) and sfcproxyrestore(1M), when possible, this hardware function is used.
The employment pattern that incorporates (shift to an equivalent state), creates a snapshot by separation in high-speed backup restoration of the GFS Shared File System which used sfcproxyop(1M) and sfcproxyrestore(1M), and performs backup restoration.
An extent is a group of contiguous data blocks of the file system that are handled as a single unit. Each extent is defined by its start block and length.
The information data of file include data that allocate a data area beforehand to a file, extend a data area in a certain proportion, and guarantee data area continuity.
The extent base is a method of managing file system data based on extents.
One of the interfaces using optical fibre cables as transmission media. In a cluster system, a fibre channel is used to connect a array disk and the computer.
File data is the data part of a file.
Property information used to specify allocation from a specific file data partition and to place a block as a normal file data.
The file data area in disk partitions contains file data. It is also called data area. The file data area is allocated in the representative partition and the file data partitions.
The file data partition is a partition that consists of a file data area (data area).
The file data partitions are added to an existing file system in order to extend it. The file data partition will be specified in file data allocation attribute information.
File lock is a means for synchronization among processes that concurrently access data in the same file. There are two types of file locks: mandatory locks and advisory locks.
Software driver to control a GFS Shared File System, located in a kernel like AC for data exchange between commands and AC.
GDS(Global Disk Services).
Volume management software for protecting data by managing a disk storage device totally within a SAN environment. For details, refer to "PRIMECLUSTER Global Disk Services Configuration and Administration Guide."
A highly reliable and efficient file system utilizing the log mechanism and extent-base area management. For details, refer to "Part 2 Global File Services Local File System" in this manual.
Software which allows various operations including GFS Shared File System configuration setting and operation management using the Web screen.
GLS(Global Link Services) software provides High Reliability of network communications by duplexing network communication paths using redundant NIC(Network Interface Card)s. For details, refer to "PRIMECLUSTER Global Link Services Configuration and Administration Guide."
The i-node contains information on the file locations on the disk, on file sizes, and on the time when the respective files were last. It also includes other information for identification.
The i-node area contains an i-node. It is located in the meta-data area.
A snapshot of a master instantly created by cooperation with the OPC (One Point Copy) function provided by a ETERNUS6000 and ETERNUS3000 disk array.
Same as the update log partition
Required information is managed when employing the GFS Shared File System. Moreover, this is used in order for sfcfrmd daemon to maintain informational adjustment. The following information is included in the management partition.
1.Information on each node that constitutes the GFS Shared File System.
2.Correspondence information on common node of each GFS Shared File System, and shared device.
3.MDS arrangement information on each GFS Shared File System.
Meta-data server. Program that centrally manages the control information of a file system (meta-data). This program operates on the primary server and secondary server. Each AC communicates with the MDS and performs file operations according to the meta-data sent from the MDS.
During operation, fails the MDS layout back to the state defined when the file system was created, if the primary MDS is not running on the primary server or the secondary MDS is not running on the secondary server there ware defined when the file system was created.
Refers to the function whereby, during operation, the MDS layout is failed back to the state defined when the file system was created, if the primary MDS is not running on the primary server that was defined when the file system was created or if the secondary MDS is not running on the secondary server.
The node with the possibility that MDS performs. MDS performs in either node that it is defined as MDS node in nodes that share the file system. The default MDS node is 2 for the primary and secondary MDS.
Same as the meta-data area
Various types of management data for the file system are collectively referred to as meta-data. Meta-data includes information on free areas, i-nodes, indirect extent blocks, directories, symbolic links, data of quota and data of ACL.
The meta-data area contains meta-data. It is allocated in the representative partition.
Partition composed only of a meta-data area to use file system extension.
A virtual device with multiple access paths for the disk array unit. An access path failure will not halt operation.
A node is a computer system that runs under one operating system.
The node identifier is a universal name that is assigned to each node in the PRIMECLUSTER system. For each node, the node identifier is normally the same as the node name. However, if the node name was assigned to the standby node such as in hot standby mode, the correspondence between physical name and node name is temporarily disrupted. In this event, the node can be managed via the node identifier, which is the name for the physical device. The node identifier is displayed as each node name using PRIMECLUSTER Management View.
The node name is the name of a node that is displayed with the "uname -n" command and identified in the communication network. For details, see uname(1) in the "Solaris X Reference Manual Collection" or the online manual.
The abbreviation for One Point Copy. The hardware function of the disk array equipment (ETERNUS6000 and ETERNUS3000) of our company. It is the function which the copy of a disk can carry out within hardware at high speed. Copy operation is completed in an instant and the emulation of the completion state of a copy is carried out about a non-copied domain. It is a function indispensable in the case of the OPC employment in high-speed backup restoration of the GFS Shared File System by sfcproxyop(1M) and sfcproxyrestore(1M).
The employment pattern that performs backup restoration by the instant copy in high-speed backup restoration of the GFS Shared File System which used sfcproxyop(1M) and sfcproxyrestore(1M). When using this employment pattern, it will be the requisite that the OPC function of our company disk array equipment ETERNUS6000 and ETERNUS3000 can be used.
The partition is a physical or logical disk slice.
This information refers to the partitions that constitute the GFS Shared File System. It includes information about the size of the meta-data area, update log area, file data area (data area) and on their layout in the configuration partition.
Node on which the management server daemon of the Web-Based Admin View operates during normal operation. The primary management server uses the menus of the GFS management view to accept operation and display requests from the client and to transfer the requests to the monitored nodes.
Node on which the MDS operates during normal operation in a cluster system. This server should be set during file system registration.
Software to configure the Cluster System. For details, refer to "PRIMECLUSTER Concept Guide."
quota is a function for limiting disk space or the number of i-nodes that can be used by the user.
The representative partition is a GFS Shared File System partition that consists of meta-data, update log, and file data (data) areas. It may also consist of meta-data and file data (data) areas.
A device used to connect an RS-232C interface to a LAN interface. With this device, users
can use the Telnet function to access the TTY console function from other devices
(personal computers) connected through a LAN.
Abbreviation for Reliant(R) Monitor Software. Software that maintains high availability of user-specified resources by providing monitoring and switchover capabilities.
Round robin allocation is a file data allocation mode for allocating separate file data to multiple file data areas.
A range of node set that can share an object when a disk class is shared.
Node on which the Web-Based Admin View management server daemon operates if an error occurs on the primary management server. A secondary management server is set when the Web-Based Admin View operating environment is set. If an error occurs on the primary management server, the management server function of the Web-Based Admin View passes to the secondary management server.
Server that takes over the MDS function if an error occurs on the primary server in a cluster system. The secondary server must be set during file system registration before it can assume the MDS function when an error occurs on the primary server. The file system status is restored to assure contiguous access from each AC.
The demand from the commands or sfcfsd daemon is received and the communication way for realizing commands sfcfsd, commands sfcprmd, and communication between nodes is offered. Management of a composition database and a management partition I/O function are offered.
Employment start and employment stop of the GFS Shared File System, taking over employment (failover) of MDS, an on-line node addition and deletion, file system composition cut return in the definition state (my gray SHON), etc. offer the function which controls MDS and AC.
Starting and stop, down surveillance of the process which constitutes a GFS Shared File System is performed. When a composition process starts and executes other demons and commands, a demand is transmitted to sfcprmd and sfcprmd executes a demon and a command according to a demand.
The process which transmits a demand is called client process and the process started from sfcprmd according to a demand is called target process.
A target process notifies the situation to a client process, when it is supervised by sfcprmd and a process carries out a normal end and unusual end.
One of operation modes of a shared disk device. Applications in each node can simultaneously access a shared disk device on their own. (Note that the terms shared disk device and shared disk have different meanings. To distinguish between the two terms, a "shared disk" may be referred to as a "shared disk for simultaneous access.")
Disk device connected to multiple nodes. FDU/FUs and disk arrays are available as PRIMECLUSTER shared disk devices. Shared disk devices can be designated as shared disks or switched disks according to whether multiple nodes are allowed to access them simultaneously. (Note that the terms shared disk device and shared disk has different meanings.)
The nodes shared the data of GFS Shared File System. Up to 4 nodes are allowed.
Object set which belongs to shared disk class. Each object may also be referred to as shared class, shared group and shared disk.
A slice is an alias of a partition.
MDS node other than primary or secondary MDS node.
The super block contains information on the type of the file system, dates of creation and updating, label information, information on size and layout of the file system, and summary information on free areas.
A copy of the super block is created so that file system processing can continue even if the original super block is damaged by a block fault in the disk device.
Disk on which the operating Solaris OE is installed. The entire disk containing a slice currently operating as the file system (or swap area) of one of the following:
The update log contains the history of updates made to the file system structure. The history information is used to make a high-speed recovery of a file system possible. The area that contains the update log is called the update log area. Data in this area is overwritten cyclically.
The update log area contains update log. It is normally allocated in the representative partition. In a multi-partition configuration in which the update log area is allocated to a separate partition, the update log area is not located in the representative partition, located instead in the separate update log partition.
Update log data that was loaded into memory.
This partition consists of the update log area. It is used to allocate an update log area that is separate from the representative partition so that update log replay can be performed at high speeds.
Update log replay updates meta-data in the update log area for which an update had been interrupted. This feature is used to recover the file system in the case of failures.
Directory blocks, symbolic link path names, and indirect blocks are collectively referred to as V-data. This data is stored in the V-data area that was allocated in the meta-data area.
The V-data area contains V-data. It is allocated in the meta-data area.
Software for system operation and management using the Web screen. You can use it by installing ESF(Enhanced Support Facility).
|
Contents
Index
|