NAME
newfs_lfs —
construct a new LFS file
system
SYNOPSIS
newfs_lfs |
[newfs_lfs-options]
special |
DESCRIPTION
newfs_lfs builds a log-structured file system on the specified
special device basing its defaults on the information in the disk label.
Before running
newfs_lfs the disk must be labeled using
disklabel(8), the proper
fstype is 4.4LFS. Reasonable values for the
fsize
,
bsize
, and
sgs
fields are
1024, 8192, and 7 respectively.
The following options define the general layout policies.
-
-
- -A
- Attempt to compute the appropriate segment size using the
formula 4 * bandwidth * access time. The disk is tested
for twenty seconds to discover its bandwidth and seek time.
-
-
- -B
logical-segment-size
- The logical segment size of the file system in bytes. If
not specified, the segment size is computed by left-shifting the partition
label's block size by the amount indicated in the partition table's
segshift. If the disklabel indicates a zero block size or segment shift, a
compile-time default segment size of 1M is used.
-
-
- -b
block-size
- The block size of the file system in bytes. If not
specified, the block size is taken from the partition label, or if the
partition label indicates 0, a compile-time default of 8K is used.
-
-
- -F
- Force creation of an LFS even on a partition labeled as
another type. newfs_lfs will use compile-time default
values for block and fragment size, and segment shift, unless these are
overridden by command-line flags.
-
-
- -f
fragment-size
- The fragment size of the file system in bytes. If not
specified, the fragment size is taken from the partition label, or if the
partition label indicates 0, a compile-time default of 1K is used.
-
-
- -I
interleave
- Specify the interleave between segments. The default is
zero.
-
-
- -i
- The size of an inode block, in bytes. The default is to use
the same size as a fragment, or in a v1 filesystem, the same size as a
data block.
-
-
- -L
- Create a log-structured file system (LFS). This is the
default, and this option is provided for compatibility only.
-
-
- -M
nsegs
- Specify lfs_minfreeseg, the number of
segments left out of the amount allocated to user data. A higher number
increases cleaner performance, while a lower number gives more usable
space. The default is based on the size of the filesystem, either 5% of
the total number of segments or 20 segments, whichever is larger.
-
-
- -m
free space %
- The percentage of space reserved from normal users; the
minimum free space threshold. The default value used is 10%.
-
-
- -N
- Do not actually create the filesystem.
-
-
- -O
offset
- Start the first segment this many sectors from the
beginning of the partition. The default is zero.
-
-
- -R
nsegs
- Specify lfs_resvseg, the number of
segments set aside for the exclusive use of the cleaner. A larger figure
reduces the likelihood of running out of clean segments, but if
lfs_resvseg is too close to
lfs_minfreeseg, the cleaner will run without ceasing
when the filesystem becomes close to full. The default is the larger of 15
or the quantity lfs_minfreeseg / 2 + 1 .
-
-
- -r
ident
- For a v2 filesystem, specify the roll-forward identifier
for the filesystem. This identifier, a 32-bit numeric quantity, should be
different from that of any LFS that may previously have existed on the
same disk. By default the identifier is chosen at random.
-
-
- -s
size
- The size of the file system in sectors.
-
-
- -v
version
- Make a filesystem with the specified disk layout version.
Valid options are 1 or 2 (the default). Note, however,
that LFS version 1 is deprecated, and 64-bit volumes may not be version
1.
-
-
- -w
bitwidth
- Create a filesystem using the specified bit width: 32 or
64. 32-bit LFS volumes are limited to 2TB in size; 64-bit LFS volumes are
not, but incur additional overhead from having larger metadata structures.
The default is 32-bit for volumes less than 1 TB, and 64-bit for larger
volumes.
SEE ALSO
disktab(5),
disklabel(8),
diskpart(8),
dumplfs(8)
M. Seltzer, K.
Bostic, M. McKusick, and C.
Staelin, An Implementation of a Log-Structured File
System for UNIX, Proceedings of the Winter 1993 USENIX
Conference, pp. 315-331,
January 25-29, 1993.
J. Matthews, D.
Roselli, A. Costello, R.
Wang, and T. Anderson,
Improving the Performance of Log-Structured File Systems
with Adaptive Methods, Proceedings of the Sixteenth ACM
SOSP, October 1997.
HISTORY
A
newlfs command appeared in
4.4BSD,
and was renamed to
newfs_lfs for
NetBSD
1.4.