[ILUG] [OT] Cascading switches

[Paul Jakma]

On Tue, 14 Aug 2001, Conor Daly wrote:

> I'd be inclined to think that could be handled quite simply by the use of a
> "live" flag.  Machine A comes up and looks for machine B.  If it doesn't
> find it, it echoes its hostname into a "live-machine" file, NFS exports /home
> and starts serving.  Now machine B comes up, looks for machine A, finds it
> and reads the "live-machine" file.  B then does an rsync on machine A's
> /home, looks for any "wrong way around" files and tries to push them up to
> A.  The difficulty arises when a file that was newer on machine B before the
> reboot gets modified on A before B comes back up.

alternatively, buy a cheap disk case (eg a 3 disker from
acme-technology.co.yuk) and manually failover. (you could have a
/etc/fstab.master and fstab.client and cp them appropriately. even
have a script).

> Conor

--paulj