[SOLVED] Switch redundancy for cluster network?

C

CZappe

Active Member
Jan 25, 2021
36
12
28
Santa Fe, NM, USA
www.bti-usa.com
Hey all,
I'm setting up my first Proxmox cluster environment after administering an environment of standalone nodes with shared NFS storage for several years. Considering the importance of a continuous, low-latency network connection between clustered nodes to establish and maintain quorum, I'm thinking about deploying a set of redundant switches, dedicated to this task. Our cluster will be small (no more than 7 nodes) so we are hoping to find an economical option that doesn't over-deliver switching capacity but still provides robust failover in the event that one switch goes offline. Our nodes will have pairs of 1GbE NICs that we can dedicate to this cluster backbone.

Has anyone else set up redundant switches for their cluster quorum? Any recommendations on hardware and switch protocols that might fit the bill for a small but fault-tolerant cluster configuration?

Thanks!
 
The following chapter of the documentation might be useful to you: Redundant Corosync Links

You can either configure multiple network devices for the cluster to use upon cluster creation via the Web UI, or add redundant links to the cluster via the configuration file as explained in the documentation. When editing the corosync.conf file, it is especially important to follow the documentation very closely to avoid problems.

As far as hardware is concerned, regular 1G switches are more than enough - Corosync does not need high bandwidth but rather very low latency.
 
Last edited:
Thank you for the feedback! The documentation on redundant corosync link configuration is exactly what I was looking for. We'll go ahead and put a couple of 1G managed switches into service as our backbone and follow this guide closely to get the redundancy configured. Much appreciated!
 
You must log in or register to reply here.
Top Bottom