Re: [Gluster-devel] how do I known afr working ?

[Anand Avati]

Albert,
 first of all 10 replicas of namespace is really too high. I would suggest
2, max 3.
with unify, a given file should exist on only 1 of its subvolumes, but for
afr it should exist in all subvolumes. so in your setup, you should have the
given filename exist in all namespace directories, with 0 byte filesize,
whereas should exist in only one unify's subvolume with the complete
filesize.

thanks,
avati

2007/11/23, Albert Shih <address@hidden>:
>
> Hi all
>
> I've just make a afr volume for the namespace of my glusterfs.
>
> How do I known the afr workin ?
>
> I've ask this because I've use one partition on every node of my cluster
> for namespace.
>
> On each client I make
>
> volume ns
>   type cluster/afr
>   subvolumes node-ns-1....node-ns-18
> end-volume
>
> (I think it's enougth raid ;-) )
>
> and
>
> volume unify
>   type cluster/unify
>     subvolumes node-1....node-18
>     option scheduler rr
>     option namespace ns
> end-volume
>
> Well it's working....I've just put one file on my unify-glusterfs. It seem
> (maybe with scheduler rr) the file is on (physicaly) the node-2. And on
> node-2 (physicaly) I can see on partition use by the namespace a file with
> the same name. But.....on other node I can see anything on the partition
> use bye afr/namespace.
>
> It's that normal ?
>
> Regards.
>
>
> NB: Sorry for my english....
>
> --
> Albert SHIH
> Observatoire de Paris Meudon
> SIO batiment 15
> Heure local/Local time:
> Jeu 22 nov 2007 20:28:25 CET
>
>
> _______________________________________________
> Gluster-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>



-- 
It always takes longer than you expect, even when you take into account
Hofstadter's Law.

-- Hofstadter's Law