[OpenAFS] R/W replication

[Nathan Neulinger]

Dirk Heinrichs wrote:
> 
> Nathan Neulinger wrote:
> >
> > Dirk Heinrichs wrote:
> > >
> > > Derek Atkins wrote:
> > > >
> > > The reason I asked was that in case of inavailability of the "master"
> > > volume, clients cannot save their changes. There could also be another
> > > solution, that in case the server of the RW volume goes down, one of
> > > it's RO replicas will automatically be switche to RW, as long as the
> > > original volume is down.
> >
> > That logic is fine so long as the volume is truly "down"... But what
> > happens if it's just inaccessible on the network. A lot of people do
> > replicas geographically distributed, in which case, that condition will
> > be relatively common.
> It could be optional, on a per volume basis.
> 
> Bye...

Right, but what do you do in that circumstance?

[client a][server b] ---/  /--- [client c][server d]

if you have rw replicates on both servers, and the link between fails -
a and b think they are fine, and c and d think they are fine. Who takes
over as master? And what tells the other server that it is no longer
master. 

What you'd probably have to do is something like forcing a quorum on
which rw volume was master, which would mean that rw volumes on small
distant networks would not be able to take over. I'm not sure that this
would benefit you much though.

Maybe if you were to distinguish between inaccessible and down somehow.
Maybe by requiring all the servers to be in constant contact for
negotiating stuff like that. 

-- Nathan

------------------------------------------------------------
Nathan Neulinger                       EMail:  nneul@umr.edu
University of Missouri - Rolla         Phone: (573) 341-4841
CIS - Systems Programming                Fax: (573) 341-4216