I followed your interesting article "Distributed Storage Across Four Storage Nodes With GlusterFS On Debian Lenny
You stated at the beginning: "...Please note that this kind of storage (distributed storage) doesn't provide any high-availability features, as would be the case with replicated storage..."
The question is exactly for this purpose: how can I implement this configuration with high-availability feature?
The doubt I have is: I can use many systems like in your example, all with Raid/Mirrors, but if one of these systems crashes, I loose the data on that system, right?
How can this functionallity be implemented? Is there a feature already good to be integrated with GlusterFS or it is something to implement 'manually' (i.e. drbd)?
Please, let us know.
Thanks and congratulations for your great work.