Re: [Bug-tar] 1.15.91 file name too long to be stored in a GNU multivolume header, truncated

[Sergey Poznyakoff]

mark jackson <address@hidden> wrote:

> the interesting thing comes now when i compare the original-directories
> and the new created directories with
> 
> "diff -r <original-directory> <new created directory>"
> 
> there is NO difference; so it seems that the backup worked perfect;
> so why are these errormessages?
> 

These are not error messages, only warnings.

When creating multivolume backups it often happens that a file
needs to be split between two volumes. In this case GNU tar
stores the file name along with some additional data at the
beginning of the next volume in a special block called volume
header. In GNU format this header can accomodate file names 
up to 100 characters long; if the actual name is longer than
that, tar truncates it and stores only the first 100 characters
in the header. This is what

"file name too long to be stored in a GNU multivolume header, truncated"

diagnostics means.

When restoring from a multivolume backup, tar uses volume headers
to verify that the user is feeding archive volumes in the right order. This
is achieved by comparing the file name and starting byte offset stored
in the next volume with the name and offset of the file being extracted.
If tar notices that the next volume contains a truncated name and the
offset of that volume matches, it continues extracting the file but
issues the warning

tar: `foo' is possibly continued on this volume: header contains truncated name

The purpose of this warning is to draw user's attention to the fact that
tar is not 100% sure this is the right volume, since it was only able to
compare first 100 bytes of the file name and the byte offsets.

Regards,
Sergey