equinix-labs / terraform-equinix-metal-nutanix-cluster

Nutanix Cluster on Equinix Metal
https://deploy.equinix.com/labs/terraform-equinix-metal-nutanix-cluster/
Apache License 2.0
2 stars 7 forks source link

Cluster migration example should not need "ip route add" steps #106

Open displague opened 3 months ago

displague commented 3 months ago

The examples/cluster-migration example currently has the user run the following command and a similar one on the two clusters. This command is only run against CVIM.

    ssh -L 9440:$nutanix_cluster1_cvim_ip_address:9440 \
        -L 19440:$nutanix_cluster1_prism_central_ip_address:9440 \
        -i $nutanix_cluster1_ssh_private_key \
        -J root@$nutanix_cluster1_bastion_public_ip \
        admin@$nutanix_cluster1_cvim_ip_address
    sudo ip route add 192.168.100.0/22 via 192.168.96.1

Ideally, all VMs and nodes in both clusters would reach nodes and VMs without additional steps.

displague commented 3 months ago

I think this may be a result of creating two VRF IP Reservations rather than a single reservation. The clusters each believe their interface has access to the whole /21, but the VRF is limiting their ARP traffic to the /22 (matching their IP reservation).

The simplest remedy may be to have both clusters use the same VRF reservation (not implemented), seeing the /21 as their metal network (currently implemented), while they use /22 as their DHCP range (currently implemented).