lrzip(1) lrzip(1)

lrzip - a large-file compression program

lrzip [OPTIONS] <file>
lrzip -d [OPTIONS] <file>
lrunzip [OPTIONS] <file>
lrzcat [OPTIONS] <file>
lrztar [lrzip options] <directory>
lrztar -d [lrzip options] <directory>
lrzuntar [lrzip options] <directory>
lrz [lrz options] <directory>
LRZIP=NOCONFIG [lrzip|lrunzip] [OPTIONS] <file>

LRZIP is a file compression program designed to do particularly well on very large files containing long distance redundancy. lrztar is a wrapper for LRZIP to simplify compression and decompression of directories.

Here is a summary of the options to lrzip.

General options:
 -c, --check             check integrity of file written on decompression
 -d, --decompress        decompress
 -e, --encrypt[=password] password protected sha512/aes128 encryption on compression
 -h, -?, --help          show help
 -H, --hash              display md5 hash integrity information
 -i, --info              show compressed file information
 -q, --quiet             don't show compression progress
 -Q, --very-quiet        don't show any output
 -r, --recursive         operate recursively on directories
 -t, --test              test compressed file integrity
 -v[v], --verbose        Increase verbosity
 -V, --version           show version
Options affecting output:
 -D, --delete            delete existing files
 -f, --force             force overwrite of any existing files
 -k, --keep-broken       keep broken or damaged output files
 -o, --outfile filename  specify the output file name and/or path
 -O, --outdir directory  specify the output directory when -o is not used
 -S, --suffix suffix     specify compressed suffix (default '.lrz')
Options affecting compression:
 -b, --bzip2             bzip2 compression
 -g, --gzip              gzip compression using zlib
 -l, --lzo               lzo compression (ultra fast)
 -n, --no-compress       no backend compression - prepare for other compressor
 -z, --zpaq              zpaq compression (best, extreme compression, extremely slow)
Low level options:
 -L, --level level       set lzma/bzip2/gzip compression level (1-9, default 7)
 -N, --nice-level value  Set nice value to value (default 19)
 -p, --threads value     Set processor count to override number of threads
 -m, --maxram size       Set maximum available ram in hundreds of MB
                         overrides detected amount of available ram
 -T, --threshold         Disable LZ4 compressibility testing
 -U, --unlimited         Use unlimited window size beyond ramsize (potentially much slower)
 -w, --window size       maximum compression window in hundreds of MB
                         default chosen by heuristic dependent on ram and chosen compression
LRZIP=NOCONFIG environment variable setting can be used to bypass lrzip.conf.
TMP environment variable will be used for storage of temporary files when needed.
TMPDIR may also be stored in lrzip.conf file.
If no filenames or "-" is specified, stdin/out will be used.

This option enables integrity checking of the file written to disk on decompression. All decompression is tested internally in lrzip with either crc32 or md5 hash checking depending on the version of the archive already. However the file written to disk may be corrupted for other reasons to do with other userspace problems such as faulty library versions, drivers, hardware failure and so on. Enabling this option will make lrzip perform an md5 hash check on the file that's written to disk. When the archive has the md5 value stored in it, it is compared to this. Otherwise it is compared to the value calculated during decompression. This offers an extra guarantee that the file written is the same as the original archived.
Decompress. If this option is not used then lrzip looks at the name used to launch the program. If it contains the string "lrunzip" then the -d option is automatically set. If it contains the string "lrzcat" then the -d -o - options are automatically set.
Encrypt. This option enables high grade password encryption using a combination of multiply sha512 hashed password, random salt and aes128 CBC encryption. Passwords up to 500 characters long are supported, and the encryption mechanism used virtually guarantees that the same file created with the same password will never be the same. Furthermore, the password hashing is increased according to the date the file is encrypted, increasing the number of CPU cycles required for each password attempt in accordance with Moore's law, thus making the difficulty of attempting brute force attacks proportional to the power of modern computers.
Print an options summary page
This shows the md5 hash value calculated on compressing or decompressing an lrzip archive. By default all compression has the md5 value calculated and stored in all archives since version 0.560. On decompression, when an md5 value has been found, it will be calculated and used for integrity checking. If the md5 value is not stored in the archive, it will not be calculated unless explicitly specified with this option, or check integrity (see below) has been requested.
This shows information about a compressed file. It shows the compressed size, the decompressed size, the compression ratio, what compression was used and what hash checking will be used for internal integrity checking. Note that the compression mode is detected from the first block only and it will show no compression used if the first block was incompressible, even if later blocks were compressible. If verbose options -v or -vv are added, a breakdown of all the internal blocks and progressively more information pertaining to them will also be shown.
If this option is specified then lrzip will not show the percentage progress while compressing. Note that compression happens in bursts with lzma compression which is the default compression. This means that it will progress very rapidly for short periods and then stop for long periods.
If this option is specified then lrzip will not show any output to the console except for error messages.
If this option is specified, lrzip will recursively enter the directories specified, compressing or decompressing every file individually in the same directory. Note for better compression it is recommended to instead combine files in a tar file rather than compress them separately, either manually or with the lrztar helper.
This tests the compressed file integrity. It does this by decompressing it to a temporary file and then deleting it.
Increases verbosity. -vv will print more messages than -v.
Print the lrzip version number

If this option is specified then lrzip will delete the source file after successful compression or decompression. When this option is not specified then the source files are not deleted.
If this option is not specified (Default) then lrzip will not overwrite any existing files. If you set this option then rzip will silently overwrite any files as needed.
This option will keep broken or damaged files instead of deleting them. When compression or decompression is interrupted either by user or error, or a file decompressed fails an integrity check, it is normally deleted by LRZIP.
Set the output file name. If this option is not set then the output file name is chosen based on the input name and the suffix. The -o option cannot be used if more than one file name is specified on the command line.
Set the output directory for the default filename. This option cannot be combined with -o.
Set the compression suffix. The default is '.lrz'.

Bzip2 compression. Uses bzip2 compression for the 2nd stage, much like the original rzip does.
Gzip compression. Uses gzip compression for the 2nd stage. Uses libz compress and uncompress functions.
LZO Compression. If this option is set then lrzip will use the ultra fast lzo compression algorithm for the 2nd stage. This mode of compression gives bzip2 like compression at the speed it would normally take to simply copy the file, giving excellent compression/time value.
No 2nd stage compression. If this option is set then lrzip will only perform the long distance redundancy 1st stage compression. While this does not compress any faster than LZO compression, it produces a smaller file that then responds better to further compression (by eg another application), also reducing the compression time substantially.
ZPAQ compression. Uses ZPAQ compression which is from the PAQ family of compressors known for having some of the highest compression ratios possible but at the cost of being extremely slow on both compress and decompress (4x slower than lzma which is the default).

Set the compression level from 1 to 9. The default is to use level 7, which gives good all round compression. The compression level is also strongly related to how much memory lrzip uses. See the -w option for details.
The default nice value is 19. This option can be used to set the priority scheduling for the lrzip backup or decompression. Valid nice values are from -20 to 19. Note this does NOT speed up or slow down compression.
Set the number of processor count to determine the number of threads to run. Normally lrzip will scale according to the number of CPUs it detects. Using this will override the value in case you wish to use less CPUs to either decrease the load on your machine, or to improve compression. Setting it to 1 will maximise compression but will not attempt to use more than one CPU.
Disables the LZ4 compressibility threshold testing when a slower compression back-end is used. LZ4 testing is normally performed for the slower back-end compression of LZMA and ZPAQ. The reasoning is that if it is completely incompressible by LZ4 then it will also be incompressible by them. Thus if a block fails to be compressed by the very fast LZ4, lrzip will not attempt to compress that block with the slower compressor, thereby saving time. If this option is enabled, it will bypass the LZ4 testing and attempt to compress each block regardless.
Unlimited window size. If this option is set, and the file being compressed does not fit into the available ram, lrzip will use a moving second buffer as a "sliding mmap" which emulates having infinite ram. This will provide the most possible compression in the first rzip stage which can improve the compression of ultra large files when they're bigger than the available ram. However it runs progressively slower the larger the difference between ram and the file size, so is best reserved for when the smallest possible size is desired on a very large file, and the time taken is not important.
Set the maximum allowable compression window size to n in hundreds of megabytes. This is the amount of memory lrzip will search during its first stage of pre-compression and is the main thing that will determine how much benefit lrzip will provide over ordinary compression with the 2nd stage algorithm. If not set (recommended), the value chosen will be determined by an internal heuristic in lrzip which uses the most memory that is reasonable, without any hard upper limit. It is limited to 2GB on 32bit machines. lrzip will always reduce the window size to the biggest it can be without running out of memory.

"make install" or just install lrzip somewhere in your search path.

LRZIP operates in two stages. The first stage finds and encodes large chunks of duplicated data over potentially very long distances in the input file. The second stage is to use a compression algorithm to compress the output of the first stage. The compression algorithm can be chosen to be optimised for extreme size (zpaq), size (lzma - default), speed (lzo), legacy (bzip2 or gzip) or can be omitted entirely doing only the first stage. A one stage only compressed file can almost always improve both the compression size and speed done by a subsequent compression program.

The key difference between lrzip and other well known compression algorithms is its ability to take advantage of very long distance redundancy. The well known deflate algorithm used in gzip uses a maximum history buffer of 32k. The block sorting algorithm used in bzip2 is limited to 900k of history. The history buffer in lrzip can be any size long, not even limited by available ram.

It is quite common these days to need to compress files that contain long distance redundancies. For example, when compressing a set of home directories several users might have copies of the same file, or of quite similar files. It is also common to have a single file that contains large duplicated chunks over long distances, such as pdf files containing repeated copies of the same image. Most compression programs won't be able to take advantage of this redundancy, and thus might achieve a much lower compression ratio than lrzip can achieve.

LRZIP recognises a configuration file that contains default settings. This configuration is searched for in the current directory, /etc/lrzip, and $HOME/.lrzip. The configuration filename must be lrzip.conf.

By default, lrzip will search for and use a configuration file, lrzip.conf. If the user wishes to bypass the file, a startup ENV variable may be set.
LRZIP = NOCONFIG [lrzip|lrunzip] [OPTIONS] <file>
which will force lrzip to ignore the configuration file.

The ideas behind rzip were first implemented in 1998 while I was working on rsync. That version was too slow to be practical, and was replaced by this version in 2003. LRZIP was created by the desire to have better compression and/or speed by Con Kolivas on blending the lzma and lzo compression algorithms with the rzip first stage, and extending the compression windows to scale with increasing ram sizes.

Nil known.

lrzip.conf(5), lrunzip(1), lrzcat(1), lrztar(1), lrzuntar(1), lrz(1), bzip2(1), gzip(1), lzop(1), rzip(1), zip(1)

Exit status is normally 0; if an error occurs, exit status is 1, usage errors is 2.

lrzip is being extensively bastardised from rzip by Con Kolivas.
rzip was written by Andrew Tridgell.
lzma was written by Igor Pavlov.
lzo was written by Markus Oberhumer.
zpaq was written by Matt Mahoney.
Peter Hyman added informational output, updated LZMA SDK, and added lzma multi-threading capabilities.

If you wish to report a problem, or make a suggestion, then please consult the git repository at:

lrzip is released under the GNU General Public License version 2. Please see the file COPYING for license details.

February 2022