Re: [Vrs-development] reliability of the repository

[Eric Altendorf]

On Friday 03 May 2002 16:43, address@hidden wrote:
> Hi
>
> Since we are thinking of a distributed file system, how are we planning to
> sync all the blocks which are spread around the VRS cluster? Suppose one of
> the remote blocks get corrupted, what happens next?

We'll have to assume that blocks are mirrored on multiple hosts, and in the 
case of a disagreement (which should be periodically tested), they will have 
to vote.  It's more involved than that obviously, but that's the idea.

> According to the Documentation , the repository on the LDS serve as minor
> block storage. What happens if and when the LDS wants to go offline? What
> happens if the LDS storing some of the blocks goes down abruptly?

Again, we have to assume that we're mirroring these blocks.  When a node 
(call it X) goes down, the other nodes which also store the blocks X stored 
will have to detect this, and replicate those blocks onto other nodes, so we 
always have several nodes online storing the blocks in question.

And these are just the easy questions... :-)

Eric

-- 
"First they ignore you.  Then they laugh at you.
 Then they fight you.  And then you win."             -Gandhi