PRIMECLUSTER Global File Services Configuration and Administration Guide 4.2 (Solaris(TM) Operating Environment)

F.1.4 mkfs_sfcfs(1M) Constructs the GFS Shared File System

[SYNOPSIS]

mkfs -F sfcfs [ generic_options ] [ -o specific_options ] raw_device_file

[DESCRIPTION]

The mkfs_sfcfs command builds the GFS Shared File System on the disks that were specified.

Once the GFS Shared File System has been created, its entry is added to file system configure information in a management partition. Thereafter, file system configure information in a management partition is used to manage the GFS Shared File System.

[OPTIONS]

generic_options

generic_options is supported by a general-purpose mkfs(1M).

See the options of mkfs(1M) for details.

-o specific_options

Allows options specific to GFS Shared File System to be specified. Insert a comma delimiter between options.

N

Displays parameters used for creating a file system. Does not actually create a file system.

clrmeta

A meta-data area is initialized altogether.

data=raw_device_file

Specifies the character-special devices for a file data partition. If a file system is built with multiple partitions that are used as file data partitions, specify multiple entries for this parameter.

dataopt=y|n

Specifies whether a representative partition includes a file data area if file data partitions are specified with the data sub-option. If y is specified, the representative partition does not include a file data area. If n is specified, the representative partition includes a file data area.

The default is n.

If no file data partitions are specified, n is set, regardless of whether this parameter is specified.

free=n

Specifies the percentage of the file data area that only a super user is allowed to use.

The default is 10%.

A value can be specified in the range from 0 to 99.

force

Specify the force option to re-create a file system previously registered in file system configure information in a management partition when its configuration is unchanged. If, however, the target file system is being mounted, specifying this option results in an error.

log=raw_device_file

Specifies a character-special device for an update log partition.

The portion of an update log partition not used, as an update log area becomes an unused area. Specify a partition that is the same size as the update log area.

logsz=n

Specifies the size of the update log area. The unit is megabytes.

The default is about 1% of the file system size. When maxdsz has been specified, the default is about 1% of the maxdsz value. If the default value is smaller than 5 megabytes, 5 megabytes is set. If the default value is greater than 50 megabytes, 50 megabytes is set.

A value can be specified in the range from 5 to 100.

maxdsz=n

Specifies the maximum size of the file data areas of each file system that includes additional file data partitions. The unit is megabytes. Based on the value specified for this parameter, the size of meta-data and the update log area size are set to proper values after taking into account the additional file data partitions.

If this option is not specified, the sizes of meta-data area and the update log area are determined based on the file system size of the specified partition.

A value greater than 32 terabyte (33,554,432 megabytes) cannot be specified.

maxnode=n

Specifies the maximum number of nodes, including any anticipated in future expansion with sfcnode(1M), that share a file system constructed on disks.

The default is 16.

A value can be specified in the range from 1 to 32.

maxvol=n

Specifies the maximum number of partitions, including any anticipated in future expansion with sfcadd(1M), that constitute the file system constructed on disks.

The default is 16.

A value can be specified in the range from 1 to 32

mdsnode=all|num

Specifies the number of MDS placing nodes. MDS placing nodes are selected from head of the host names specified with node option. When all is specified, all nodes are set up as MDS placing node. If and only if all is specified, the node added to node information by sfcnode(1M) after constructing the file system is also set up as a MDS placing node.

The default is 2.

0 or less value and value bigger than the number of nodes specified with node option cannot be specified.

metasz=n

Specifies the size of the meta-data area. The unit is megabytes.

The default is about 10% of the file system size. When maxdsz has been specified, the default is set at about 10% of the maxdsz value.

The size of the meta-data area shall not exceed the size of the representative partition.

mfree=n

Specifies the percentage of the V-data area that only a super user is allowed to use.

The default is 10%.

A value can be specified in the range from 0 to 99.

nblkpext=n

Specifies an average number of blocks to be managed by one extent.

This parameter is used as a fragmentation factor for the file data area.

The default is 1.

A value can be specified in the range from 1 to 100.

nbpi=n

Specifies the byte count of data per i-node. Specify a larger value for this parameter if the meta-data area is to be saved.

The default is 8192.

A value can be specified in the range from 512 to 1,048,576.

node=host[:host],...

Registers the name of a host that will share the file system. If multiple nodes share the file system, specify multiple entries for this parameter.

Hosts whose names are to be specified with host must be able to communicate with each other.

MDS placing nodes are selected from head of the host names specified with this option. The number of MDS(meta-data server) placing nodes are specified by the mdsnode option.

Unless this option is specified, only the host name of the local node is registered. A maximum of four entries can be specified.

Use a colon delimiter to specify multiple host names. The multiplexed network path is used for communication path with AC and MDS, The host name of each node can be specified to four maximums.

opt=s|t

Selects either space or time on which to base the optimization of free block allocation.

s indicates space and t indicates time.

The default is t.

[OPERAND]

The following operand can be specified:

raw_device_file

Specifies a character-special device for the representative partition of the GFS Shared File System. A device specified for a file data partition or an update log partition cannot be specified for the representative partition.

[EXAMPLES]

# mkfs -F sfcfs -o node=host1,host2 /dev/sfdsk/gfs/rdsk/volume1

# mkfs -F sfcfs -o node=host1,host2 -o log=/dev/sfdsk/gfs/rdsk/volume2 /dev/sfdsk/gfs/rdsk/volume1

# mkfs -F sfcfs -o node=host1,host2 -o log=/dev/sfdsk/gfs/rdsk/volume2,data=/dev/sfdsk/gfs/rdsk/volume3 -o data=/dev/sfdsk/gfs/rdsk/volume4 /dev/sfdsk/gfs/rdsk/volume1

# mkfs -F sfcfs -o mdsnode=all -o node=host1,host2,host3,host4 -o log=/dev/sfdsk/gfs/rdsk/volume2 /dev/sfdsk/gfs/rdsk/volume1

# mkfs -F sfcfs -o mdsnode=3 -o node=host1,host2,host3,host4 -o log=/dev/sfdsk/gfs/rdsk/volume2 /dev/sfdsk/gfs/rdsk/volume1

[FILE]

/etc/vfstab

Environment file for the file system.

[SEE ALSO]

sfcadd(1M), sfcnewfs(1M), sfcnode(1M), sfcsetup(1M), sfctune(1M).

fsck(1M), mkfs(1M), vfstab(4), and attributes(5) in the "Solaris X Reference Manual Collection."

[NOTES]

• host is necessary to be the host name of the node which constitutes the GFS Shared File System registered into the management partition by sfcsetup(1M).
• A file system can be created only when all hosts that share the file system have started operation and GFS Shared File System has been installed. The reason is that the information for these hosts is required.
• If another option is specified after a comma (,) that follows a node option, the option will have no effect because it will be interpreted as a host name.
• In the GFS Shared File System, the data block size is fixed at 8192 bytes and cannot be altered.
• To re-create the GFS Shared File System by modifying its configuration, first delete the file system with sfcadm(1M).
• When a file system is created, no device being added can be specified. Use sfcinfo(1M) to confirm whether the device to be specified is being added.
• The file system configuration must be MDS can place in all nodes, if building the file system (if possible).
• The I/O performance of the application has influence while MDS perform meta-data area initialization after the file system start. Specify a clrmeta option to avoid this influence.
• When using a quota function, please initialize meta-data areas, with clrmeta option.