[TUTORIAL] Creating a High Availability (HA) Shared Storage of a Two-Node DRBD-Based Cluster with RDMA-Based Networking

[Oleksandr.V]

With the falling price of network cards, there was a desire to build a two-node cluster with two points of failure, with an NVME disk on each node for shared storage. This was supposed to be a cluster option for home or a small business. The choice fell on drbd storage with rdma. ​

https://vidomenko-it.github.io/blog/post-1/

 

[RolandK]

it seems there is no mention about the pve clustering - just setting up the drbd mirroring. so i have difficulties in understanding how this can work, as we need three nodes minimum with proxmox pve.

 

[waltar]

You use any 2 nodes (pve internal or external) for redundant storage representation and use all (or all others) pve nodes just as clients for vm/lxc images.