[Gluster-devel] Question about disk space

[Daniel van Ham Colchete]

On 6/3/07, Brandon Lamb <brandonlamb at gmail.com> wrote:
>
> Hopefully one of the devs can answer this.
>
> I just setup a 2 server cluster. server1 has 144 gigs of free space,
> server2 has 78 gigs.
>
> I have both set to do the AFR thing and told the config to replicate *
> (all files right?) with 2 copies.
>
> So what happens when I get to 80 gigs, will i start getting errors or
> will glusterfs just stop creating replications on server2 since it
> doesnt have enough space?


I'm not a Gluster developer, but I might have the answer for you.

If you "df -h" on the glusterfs mounted dir you'll see that the total space
of the mount point is the lower total space between the AFR bricks. So you
will not get any gluster errors, just the usual ENOSPC errno: "The device
containing the file referred has no room for the data".Usually both AFR
bricks should have the same available disk space, but not necessarily the
same filesystem or the same geometry (as in DRBD).

Now that I think about it, GlusterFS is also the perfect solution for
testing different local filesystems performance for each specific case: if
you can't decide whether ReiserFS or XFS is better for storing Maildir, try
both! Do your tests with your real data as long as you want and format the
losing side later. The more I think about GlusterFS's design, the more I
love it.

Best regards,
Daniel Colchete