[Gluster-users] GlusterFS on a two-node setup

[Brian Candler]

On Sun, May 20, 2012 at 06:09:47PM -0400, David Coulson wrote:
> >I might have to look at DRBD more carefully, but I do not think it
> >fits my needs: I need both nodes to be working (and thus doing
> >I/O) at the same time. These are basically number crunching nodes
> >and data needs to be accessible from both nodes (e.g., some jobs
> >will use MPI over the CPUs/cores of both nodes ---assuming both
> >nodes are up, of course ;-).
> DRBD will let you do read/write on both nodes, but it requires a
> clustered filesystem such as GFS2 or OCFS2 on top of it.

Or more simply, with DRBD you can just split your space into two and mirror
half in each direction.

Server A: 300GB volume replicated to server B
Server B: 300GB volume replicated to server A

If one server dies, both volumes can be mounted on the same server.

But if you need to access the *same* data on both nodes simultaneously, then
you'd end up having to share it using something like NFS, in which case you
may as well use glusterfs in the first place.

> You are
> also limited to a max of two nodes.

Well, DRBD has to be configured in pairs (e.g. you can have one volume A->B,
another B->C, another C->A).  This is what Ganeti gives you: each VM has its
own DRBD on top of two LVM volumes on two nodes of your choice.

Ganeti is primarily of benefit when doing a cluster of VMs.

You can also do VMs with an image files stored on a gluster replicated
volume.  I've not tested this, but I'm told that gluster 3.2 will lock the
whole file while healing, whereas gluster 3.3 will perform much better.

Regards,

Brian.