[TriLUG] Linux Clustering (high availability) and file systems

Timothy A. Chagnon tchagnon at nc.rr.com
Tue Apr 27 21:03:28 EDT 2004


On Tue, 2004-04-27 at 14:42, Tarus Balog wrote:
> Okay, I want to make a particular file system highly available.
> How do Linux Clusters handle making data highly available.

This is kind of tangent and probably not at all what you want to use,
but I'll throw it in to the discussion.

On our cluster we've been playing with NBD (network block device) which
basically makes a regular file or block device (partition) on machine1
look like a block device on machine2.  For our setup this looks like:
machine1:/dev/nd0 == machine2:/dev/hda5

Of course this just skips the whole NFS thing.  The kernel module is
under unsupported in the fedora kernels at least.  I don't know about
2.6 or vanilla 2.4.  You have to get the client and server programs
here: http://nbd.sourceforge.net/ and making them use large file support
is a bit tricky, but it works!  The client/server programs really need
some more development in the disconnect handling area.

Like I said, it's probably not what you want, but it's an interesting
concept.
-- 
Timothy A. Chagnon <tchagnon at nc.rr.com>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://www.trilug.org/pipermail/trilug/attachments/20040427/202d55c3/attachment.pgp>


More information about the TriLUG mailing list