[Tfug] Communications reliability

[Bexley Hall]

Hi,

--- "Eric M. Gearhart" <eric at nixwizard.net> wrote:

> I don't have much to add to this, other than wow
> that sounds *really* cool. Basically the nodes on
> the cluster are assumed to be unreliable, and if one
> goes down the workload that specific node was
> assigned would just be dynamically reassigned. 

No, my goals are different.  I am just saying "if
a machine is down or otherwise not accessible to 
some *other* machine needing its services, then those
services are just not available to that "client".

The "issue" is how this should be handled so that the
user isn't bombarded with technobabble messages
(about "host is down", etc.).

Imagine a computer in which portions of the *memory*
can suddenly stop working.  Or, a peripheral going
"off-line".  (bad examples)  You still want to be
able to use whatever aspects of thatt machine that
do NOT rely on those "unavailable resources".

--don


      ____________________________________________________________________________________
Never miss a thing.  Make Yahoo your home page. 
http://www.yahoo.com/r/hs