[Gluster-users] Load balanced VIP

[Kaleb Keithley]

----- Original Message -----
> From: "Anthony Valentine" <anthony.valentine at bldr.com>
> 
> I am working on implementing my first Gluster/Ganesha NFS setup and I am
> flowing this guide:
> http://blog.gluster.org/2015/10/linux-scale-out-nfsv4-using-nfs-ganesha-and-glusterfs-one-step-at-a-time 
> 
> Everything is working fine. I’ve got Gluster and Ganesha NFS working and I
> have VIPs on each node and it is failing over fine, if a little slowly.
> However, the VIPs don’t behave as I was expecting.
> 
> I was expecting a single VIP that clients could connect to that would load
> balance amongst all the active nodes, instead of having a VIP on each node.
> Is it possible to configure it to behave this way? I’m happy adding a proxy
> layer, such as an F5 or HAProxy, however I want to make sure that Ganesha
> doesn’t handle this on its own before I head in that direction. Also, if it
> does require a proxy, is there a particular product that is known to work
> well and is there a guide for setting that up somewhere?
>

The Gluster/Ganesha HA provides a simple Active/Active HA.

I'm not aware that Pacemaker has any load-balancing capability in the IPaddr RA.

I think you will have to implement that yourself.

--

Kaleb