At the University of Georgia we used IBM's DCE/DFS for our email/web cluster.
That is really the only distributed file system that I've used that actually
works as advertised. But, make sure you keep a dev cluster around; we had
gotten to the point where there were a lot of bugs in our implementation and
we couldn't upgrade. Thus we were forced to reboot the cluster once a week
to maintain stability.
On Friday 27 June 2003 06:56 am, Joseph Mack wrote:
> mb@xxxxxxxxxxxxxxxx wrote:
> > Another question: Has anyone succesfully rolled-out a cluster of
> > real-servers using coda? (Main reason would be for the replication of
> > config files (Apache/qmail) across all real-servers) - Is this doable? Or
> > am I better of using rsync?
> Wensong talked about having done it a few years back. I assume it was a
> test setup only. There's none in production AFAIK. Distributed file systems
> are still research topics. In the beowulf mailing list, only PVFS is being
> used and not all that much (it's fragile, if one node goes down, you loose
> the whole FS - this doesn't matter a whole lot to beowulfers).
> rsync is used often for LVS's and seems to be the way to go.
Production Systems Support - Unix
Enterprise IT Services
University of Georgia
Here is your fortune:
Tuesday is the Wednesday of the rest of your life.