Content-aware splitting of tar archives

splitar splits large tar archive, writing each volume as (almost) independent tar archive, never splitting it in the middle of a file.

splitar works in a streaming fashion, never loading into the memory more than necessary.

Features

• Split tar files, generating valid tar files of limited data size.
• Reading data from stdin if file is - (writing to stdout is not possible, obviously).
• Compress filter (or any other kind) for the output.
• Optionally recreate directory entries for each new volume.
• Optionally fail when file too large is found.

Limitations

• While chunks are limited by size, it is the size of contained data, not file output.
• If the input archive contains file larger than chunk size limit, the output chunk will inevitably contains the entire file, as splitar never splits contained files. You may use option --fail-on-large-file if you want the util to fail on such a file.
• If a volume contains a hardlink or symlink to file in some previous volume, tar will refuse creating this link if the target does not exists (e.g. it was not unpacked from one of previous volumes).