Tar UTF-8 conversion error

We're occasionally seeing the following errors when using tar
under Solaris 9 and Solaris 10 (SPARC):

UTF-8 conversion failed.


invalid character in UTF-8 conversion

Clearly, there seem to be problematic characters in
the filenames.

Is the only solution to rename the files and remove the
incompatible characters?

I'm also hoping to understand the root of the problem here.

Is it accurate that tar is attempting to store the filename
in Unicode, and when it encounters a character in the filename
that cannot be stored in Unicode, tar is unable to archive the

Thanks in advance.

Stanley E. Laufer
Network Administrator
School of Library and Information Science
San Jose State University
sunmanagers mailing list

Relevant Pages

  • Re: Dateinamen Windows, UDF (DVD) bzw. ISO9660 (CD) kompatibel machen
    ... Und im Falle von Tar bekommt man u.U. bereits dann ... | 3.170 Filename Portability ... | Filenames should be constructed from the portable filename character ... | character set as octets with the most significant bit zero. ...
  • Re: Untarring / German umlaut characters
    ... create UTF-8 file names. ... did not explicitely use the format option. ... command line editing keys don't work as expected, the shells or terminal programs ignoring or misinterpreting character keys up to tools completely running berserk if they encounter a completely valid character. ... While tar tvf tarfile runs fine under certain settings, tar xvf tarfile finally files using the same settings. ...
  • Re: [newbie] cp -r across volumes - making it carry on if it fails..
    ... And WHAT command line? ... character by character, do an ls, and pass the output thrugh od -a. ... > little trick using tar etc.. ... tar and rsync will do a cp foo bar1; ...
  • Re: Versioning file system
    ... uses legal path character, but doesn't break anything. ... userspace tools will have to be taught that: ... by never actually showing .snapshot in directory listings, ... You never risked having tar ...
  • Re: CR and LF
    ... > "The character b has no effect, but is allowed for ISO C ... pages describe what *Solaris* does. ... And indeed that is the same for all unix systems. ... is strictly for compatibility with ANSI X3.159-1989 ...