next up previous contents index
Next: 10.3 The Multivolume Archive-Menu Up: 10. The MULTIBUF Program Previous: 10.1 What is MULTIBUF

10.2 How does it work?

MULTIBUF can be run in two modes, ``write-mode'' (default) and ``read-mode''. In ``write-mode'', it reads data in small blocks from stdin and writes it in bigger chunks (``superblocks'') of several blocks to the output device specified. In ``read-mode'', it does the contrary, reading big chunks of data from the device and outputs the data to stdout.

Whenever writing a big data chunk fails, MULTIBUF assumes that it has reached the end of the backup medium (e.g. end of tape) and executes one of two user-definable commands. On successful execution of this command, it optionally executes a second command and then continues writing to the device with the last block that did not fit onto the medium before. During restoration of the data, it does more or less the same, prompting the user when it cannot read a complete chunk of data. It is important to mention, that for writing and reading data, you need to use the same block sizes and number of blocks per superblock!

To improve reliability and error checking, MULTIBUF can write so-called ``sequence information headers'' to the tape. These headers tell MULTIBUF about the right order of backup media and thus prevents you from using the tapes in the wrong order. Further on, this enables MULTIBUF to distinguish between an ``end of medium'' and an ``end of file'', sparing you some unnecessary and possibly confusing dialogs. Also, this increases the number of files per backup on a tape from currently three to five, because two additional short end-markers get written. On mountable block devices MULTIBUF will write the end markers as separate small files.


next up previous contents index
Next: 10.3 The Multivolume Archive-Menu Up: 10. The MULTIBUF Program Previous: 10.1 What is MULTIBUF   Contents   Index
David Frascone
2000-10-13