seemingly corrupt downloaded gzip'd tar files

merlinblack · 02-03-2010, 04:55 AM

I'm on Ubuntu Karmic,
gzip 1.3.12
tar 1.22

I've been encountering a few downloaded tar.gz files that tar complains are corrupt. Lots actually.

Now I usually do this:

Code:

tar xvf archive.tar.gz

To extract. Tar sees the gz extension, and its as if I added "z" to the options.
However lately I've had some files where I get the error (event with the 'z' option):

Code:

tar: This does not look like a tar archive
tar: Skipping to next header
tar: Exiting with failure status due to previous errors

I and I assumed it was corrupt. I redownloaded... no good. I got suspicious when the 5th different one or so from sourceforge failed extract. A bunch from other sites nailed it home. So some experimenting...

Code:

file weird.tar.gz

Said it was a gzip file. So far so good.

Code:

gunzip weird.tar.gz

Worked. Getting closer.

Code:

tar tvf weird.tar

Whoa a list of files!

Code:

tar xvf weird.tar

Worked. Huh? Then why did the normal way not work? Even with the 'z' option added?

Code:

tar czf test.tar.gz diroffiles
tar tvf test.tar.gz

Also works.

I dunno. Perhaps tar is compiled against an older gzip library or something. So if you are getting "corrupt" gzip'd tars, try unziping and untaring as separate steps.

chrism01 · 02-03-2010, 09:55 PM

This may be relevant

Quote:

--format FORMAT
selects output archive format

v7 - Unix V7
oldgnu - GNU tar <=1.12
gnu - GNU tar 1.13
ustar - POSIX.1-1988
posix - POSIX.1-2001

-g, --listed-incremental F
create/list/extract new GNU-format incremental backup
-G, --incremental
create/list/extract old GNU-format incremental backup

http://linux.die.net/man/1/tar

AwesomeMachine · 02-03-2010, 09:57 PM

tar zxvf <file.tar.gz>
tar jxvf <file.tar.bz2>

merlinblack · 02-03-2010, 11:35 PM

Quote:

Originally Posted by AwesomeMachine

tar zxvf <file.tar.gz>
tar jxvf <file.tar.bz2>

As I mentioned in the first post, adding 'z' to the options does not help. Except when creating an archive, GNU tar will infer 'z' or 'j' from the file name, if you do not specify it explicitly.

Thanks chrism01, I'll look into that. You might be onto something there.