I have five nodes in a cluster and am about to spin up a sixth for a special case. The first five currently all use the same network (working on separating out the management interfaces from the LACP bonds so I can trunk VLANs in on the rest, but that's for another day), but the VM that will run on the sixth needs to have it's own dedicated physical network. I still want it to be a part of the cluster and quorum for ease of management, so I was wondering if there is anything wrong with having the management interface on the same network as the others, but the rest of the NIC's on a physically separate network for the VM. Also, I plan to eventually redo the other five nodes to participate in a Ceph/HA configuration with each other, but with the sixth host still just doing it's own thing while still in the cluster. Are there any glaring problems that I'm overlooking?
Quorum member with different network and storage?
- Thread starter LBX_Blackjack
- Start date
If you plan to have such VM with its own network, and it's requirements are not compatible with your actual cluster setup, it makes no sense to add that sixth node to you cluster. Just install a standalone Proxmox server and install that VM there or try to make it "fit" in your current cluster.
I can't really say if adding such sixth node to you cluster is possible or not, but even If is, it won't add much value as that special VM won't be able to be run in any other node (so no live migration, reduced resiliency) and maybe the sixth node won't be able to run any other VM. Also, having 6 nodes in your cluster will make it have at least 4 of them running to get quorum (you need more than half the amount of nodes to be up), while in your current 5 node cluster you get quorum with just 3 of them.
I can't really say if adding such sixth node to you cluster is possible or not, but even If is, it won't add much value as that special VM won't be able to be run in any other node (so no live migration, reduced resiliency) and maybe the sixth node won't be able to run any other VM. Also, having 6 nodes in your cluster will make it have at least 4 of them running to get quorum (you need more than half the amount of nodes to be up), while in your current 5 node cluster you get quorum with just 3 of them.
Yeah the only reason I wanted to add it to the cluster was to manage it from the same point, rather than having to use two portals. The reason the VM needs its own network is because it will likely saturate our primary network and I don't want it bogging down other traffic. Security is a concern as well since it's possible this second network would be exposed to malware (though that could be handled with a VLAN easily enough).