According to this page, one can let tar create a tar archive "split" into 100 Mb files:

tar -c -M --tape-length=102400 --file=disk1.tar largefile.tgz

The problem is that this command will require you to interactively give a new filename for the next file, after the first file is filled.

Anybody knows of a way to skip this interactive step, and let tar do the "splitting" automatically?

1

7 Answers

Take a look at the --new-volume-script option, which lets you replace the prompting mechanism with a different mechanism or with a generated filename. ((tar.info)Multi-Volume Archives in the tar info page.) The problem with split is that you need to cat the pieces back together to do anything, whereas a multivolume archive should be a bit more flexible.

3

You can use split for this:

tar czpvf - /path/to/archive | split -d -b 100M - tardisk

This tells tar to send the data to stdout, and split to pick it from stdin - additionally using a numeric suffix (-d), a chunk size (-b) of 100M and using 'tardisk' as the base for the resulting filenames (tardisk00, tardisk01, tardisk02, etc.).

To extract the data afterwards you can use this:

cat tardisk* | tar xzpvf -

1

Of course the best option to use is the --new-volume-script option.

But, if you know the size of the file (in this case, largefile.tgz), then you can do this also:

tar -c -M -L 102400 --file=disk1.tar --file=disk2.tar --file=disk3.tar largefile.tgz

Summary:

-c = Create
-M = multi-volume
-L 102400 = 100MB files (disk1.tar, disk2.tar, disk3.tar ...)

(For the -L, specify as many as needed so that the total sum of the tar files is larger than largefile.tgz)

If you are trying to tar a directory tree structure

it will automatically create files of size 1.1GB, if your tar is bigger in size, you can increase the number, for an example 1000 {2..1000} or you can increase the input to tape-length argument.

tar --tape-length=1048576 -cMv --file=tar_archive.{tar,tar-{2..100}} backup.tar.lzma

I've answered here with the better explanation here tar split archive

TLDR;

This command is creating 2GB chunks without the compression:

tar -cv --tape-length=2097000 --file=my_archive-{0..50}.tar file1 file2 dir3

I got it to work with the following commands:

mkdir -p ./split
rm -rf ./split/*
tar -cML 102400 -F 'cp "${TAR_ARCHIVE}" \ ./split/part_${TAR_VOLUME}.tar' \ -f split/part_1.tar large_file.tar.gz

The only problem is that part_1.tar will actually be the last file, and the others are shifted by one. I.e. part_2.tar is actually the first part, and part_k.tar is the (n - 1)^th part. Fixing this with some shell script is trivial, and left as an exercise for the reader.

The easiest non-interactive way would be -

Creating multipart tar from a large file (say bigfile)

$ tar cvf small.tar -L1024 -F 'echo small.tar-${TAR_VOLUME} >&${TAR_FD}' bigfile

Extracting multipart tar to a specific directory (say /output/dir)

$ tar xvf small.tar -F 'echo small.tar-${TAR_VOLUME} >&${TAR_FD}' -C /output/dir

You are free to select any supported compression format, block size (1024K here) and archive name (which is small.tar here).