HP-UX Reference (11i v1 05/09) - 1M System Administration Commands A-M (vol 3)

m
mkfs_vxfs(1M) mkfs_vxfs(1M)
applicable to the Version 3 and 4 disk layouts, which fix the allocation unit size at
32768. For disk layout Version 2, this is an alternate way of specifying the
number of allocation units. This option cannot be used with the
nau option.
With
ausize, the last allocation unit on the file system can be shorter than the
others. If the last allocation unit on the file system is not long enough to contain
an entire allocation unit header, the size of the resulting file system is shortened
to the end of the last complete allocation unit. This parameter may not exceed
262144 blocks.
The algorithm used to choose the default value is rather complicated, but is
intended to balance the number of allocation units (4 to 16 is a good range), the
size of the allocation units (at least 32768 blocks), and other factors.
bsize=n
n is the block size for files on the file system and represents the smallest amount
of disk space allocated to a file. n must be a power of 2 selected from the range
1024 to 8192. The default is 1024 for file systems less than 8 gigabytes, 2048 for
file systems less than 16 gigabytes, 4096 for file systems less than 32 gigabytes,
and 8192 for larger file systems.
inosize=n
n is the on-disk inode structure size for files on the file system. The valid values
are 256 and 512 bytes. The default is 256. There is usually no reason to increase
the inode size, and not using the default value can adversely affect file system
performance.
largefiles |nolargefiles
Valid only for the Version 3 and 4 disk layouts. Controls the largefiles flag for
the file system. If
largefiles is specified, the bit is set and files two giga-
bytes or larger can be created. If nolargefiles is specified, the bit is cleared
and files created on the file system are limited to less than two gigabytes. The
default is nolargefiles . See fsadm_vxfs(1M).
NOTE: Large files are supported on HP-UX 10.20 systems and above. Be careful
when implementing large file system capability. System administration utilities
such as backup may not operate correctly if they are not large-file aware.
logsize=n
n is the number of file system blocks to allocate for an activity logging area. The
minimum value for Version 2 and 3 disk layouts is 32 blocks. The minimum
value for Version 4 disk layouts is the number of blocks that make the log no less
than 256K. The maximum value for n is the number of blocks that make the log
no greater than 16384K. This means that for a bsize of 1024, 2048, 4096, or 8192
bytes the maximum value of logsize is 16384, 8192, 4096, or 2048 blocks, respec-
tively. To avoid wasting space, the default logsize is 1024 blocks for a file system
8 megabytes or larger, 128 blocks for a file system 2 megabytes or larger but less
than 8 megabytes, and 32 blocks for a file system less than 2 megabytes.
A large log provides better performance on metadata-intensive workloads. A
small log uses less space on the disk and leaves more room for file data. for
example, and NFS-intensive workload performs better with a large log; a small
floppy device requires a small log.
NOTE: The amount of virtual memory required by fsck (see fsck_vxfs(1M)) to
check a VxFS file system is proportional to the size of the log. The maximum
amount of virtual memory used is twice the size of the log. Therefore, the sum
of physical memory and swap space must be at least 32 MB to ensure that a file
system with a 16384K log can be cleaned. On small systems, take care not to
create a file system with a log larger than half the available swap space. A max-
imum log size of one third the total of memory and swap space is a good rule of
thumb (see swapinfo(1M)).
Section 1M−−488 Hewlett-Packard Company − 2 − HP-UX 11i Version 1: September 2005