Re: [Ltsp-discuss] High Availablity LTSP

[Dan S. <str...@dc...>]

On Sat, 2005-02-26 at 14:30 -0500, chr...@li... wrote:
> > If you want to make it even more bullet-proof you could have two such
> > servers, and link them together using RAID-1 over enbd
> > (enbd.sourceforge.net). So you run a network raid-1 over a physical
> > raid-5. You must only mount the filesystem on one server at the time,
> > but the other one could stand by to mount the filesystem and export it
> > (also taking over the IP? So that clients can reconnect transparently?)=
.
> >=20
> > (If you use an Oracle filesystem you could mount the filesystems on man=
y
> > servers at the same time though).
> >=20
>=20
> OCFS (oracle cluster filesystem) is not a true fielsystem.  It's only goo=
d
> for raw oracle data. RedHat has a cluster filesystem it acquired via thei=
r
> Sistina takeover.  There's another one called Luster I think.
>=20
> - christoph

Is it possible, that perhaps OCFS will present a large block device to
linux, suitable for putting something like XFS overtop of?

Thanks!