[Gluster-users] Network redundance

[Alessandro Briosi]

Il 06/04/2017 14:08, Pavel Szalbot ha scritto:
> Hi, does your stack support MC-LAG (multichassis link aggregation
> group)? If so, configure LAGs with interfaces on both switches and you
> should be fine.

Yes it does, and I thought on using it too, but then all the ports
should be redundant (not only the ones I decide)
This are 2x48 ports switches so I basically would be loosing something
like 35 ports 'cause most of the connected hosts do not have bonded
network cards.

This switches also support staking (so to see it as 1 single switch with
a special cable), which basically is pretty cool, central configuration,
bonding across the switches, etc) The problem here is that if 1 switch
goes crazy it will block the other switch too (so not redundant). This
happened to me on another installation where 1 switch blocked the whole
stack, so redundancy wasn't working as expected.

I also tought about using OSPF between the hosts though it seems rather
complicated to setup (or at least too much for a 3 host cluster)

Should also be possible with VRRP but I'm not sure how to set it up.


Alessandro