It's big file (10GB) and was created by tar in RH8, on an NFS-mounted file server.

Tried to extract or list it with the same RH8 tar, and with several much more recent versions, via NFS and SAMBA on the original server. Also on a directly-mounted ext3 partition, all with the same result.

My command is: tar -tvvf foobar.tar

And the result is... nothing at all. tar returns immediately, but apparently does nothing and issues no error messages. Nothing in /var/log/messages either. If this is tar being doubly verbose, I have to wonder what terse looks like...

Presumably this must be a simple one, but I can't see it.

Can more experienced eyes help? Or point me to another place to ask the question?

We really need to extract this archive to rebuild a broken website, so many thanks in advance for any help...

I know almost nothing about NFS, but google seems to say there is a 2Gb limit for NFSv2. Could that be part of the problem?

I would also want to have a look at the contents to make sure it is not just all zeros. A null (all zero) file gives the response you observe

Something like 'od -x bigtarfile | less' maybe. I don't remember what the layout of a tar is supposed to be but it should not be hard to find out.

NFS: The 2GB file limit is in NFS v2, but not v3, and the 10GB file WAS created via NFS after all.

tarfile content: no idea whether it's right, but it's not null - just issuing "tail foobar.tar" shows that.

But anyway, why wouldn't tar report that it's an invalid archive file?

Tar interprets a big enough block of zeros as the end-of-file marker, so if the file is all zero (or begins with enough zeros) it just assumes an empty archive. I checked.

There is a -i switch to tell tar to ignore zero blocks. Might be worth a try.

Otherwise the tar file format is described in wikipedia:

http://en.wikipedia.org/wiki/Tar_(file_format)

Try a hex dump as suggested and check you have a valid header.

Excellent!
Using -i gets me some sensible output.
Many thanks.