[Gluster-users] Distrbute AFR Storage

[Vikas Gorur]

2009/3/28 Simon Liang <simonl at bigair.net.au>:
> Is there a way to use a round-robin or something, as opposed to hash?
>
> The problem with distribute is that it does not know the disk has run out of
> space, until it's starting copying and run out of space.
> Can this be avoided?

As Sean Davis said in another mail in this thread, hashing is
deterministic. So if you keep trying with that file, you will get the
same result. This deterministic hashing makes many operations fast.

If you want "scheduling", you can try using the (slower) unify
translator instead of distribute.

Vikas
-- 
Engineer - Z Research
http://gluster.com/