Knfsd vs. old nfsd

Russell Steffen (rsteffen@ia.net)
Tue, 30 Mar 1999 11:30:12 -0800 (PST)


I was just experimenting with the kernel-based nfs daemon, and I've
noticed a disparity in it's behaviour verses the old userland nfs daemon.
I wondering if this is normal or if I'm just missing something.

This started when I wanted to do nightly backups of a WinNT server to a
Linux box. I couldn't get the SMBfs client in 2.0.36 to work reliably (and
that started last weeks thread about smbfs timeouts). I was able to get
the 2.2.x SMBfs to work very well, but I don't want to upgrade the main
server to 2.2.x just yet. So I built a SMB-to-NFS gateway on a more
disposable box. I started by mounting all of the interesting NT shares
under a single directory, and then NFS-exporting that directory.

ie. /mnt/server/
+ share1/
+ share2/
+ etc...
And just /mnt/server in /etc/exports.

On the machine doing backups, I just had to mount the /mnt/server
directory, and everything worked as planned.

Knfsd saw things differently...

Without changing anything except the nfs daemon, the client machine now
sees the share-mount points under /mnt/server, but they are empty. The
client has to explictly mount each share directory after mounting
/mnt/server. It's like knfsd doesn't like having an export that crosses
a mount point. Is that known behaviour with knfsd or am I doing something
wrong?

Russ
rsteffen@ia.net

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/