2 node Proxmox cluster + Backup server Corosync network?

Icefire

Member
Mar 20, 2022
18
0
6
44
So now we have 3x R730XD server, all with 2x25GB SFP+ and 2x10GBe.

I'm building a new High Availability network for the new office, so I want to cluster the proxmox servers have HA also on application/servers with automatic backup

What would be the best way to wire them and how to cluster them?

I read you normally do a 2 node cluster and add a PBS with a voting for quorum ?

Do I use a 10GBe Switch and aggregate in 2x10GB with only the 3 servers on it for Corosync? Would use the 2x25GB plugged on 2 Aggreate switch for HA and spare quorum link.
Does the "quorum switch" need to be standalone with nothing on it or I can just isolate the 3 server with a VLAN ? I would use a 10GB SPF+ agregation switch with SPF+ to 10GBe adapter and have the 3 server only.

Thanks
 
Last edited:
No-Single-Point-Of-Failure.png
 
Nobody have 2 nodes + PBS as 3rd device?

Considering I have 2 identical 24 drives server + 1 server with 40 drive, I was thinking it was better to do this than a 3 node with one running PBS on a VM.

I got a 10GB switch coming and will do 2x10GB/node as quorum and 1x25GB as normal use on the 2 node cluster. a second 25GB link with be "standby" as in case the first switch die. Can't run stacked switch so will try 2 switch with rSTP
 
what about storage plan ?
Server 1 have 16+24 disk which give me 2 zfs pool of 14Tb (28 TB total)
Server 2 have 24 disk with a zfs pool of 28TB 2 linux and 4 windows VM
Server 3 have 24 disk with a zfs pool of 28TB nothing on it right now.

currently total data from Server 1 and 2 have maybe only 3-4TB total. which is spread and would be easy for the Server 1 to backup.
 
Last edited:
imo, HA require Shared Storage.
Official docs says :
"High-Availability is allowed in combination with storage replication, but there may be some data loss between the last synced time and the time a node failed."

if it's ok for your use case, PBS as third qorum device is ok.