Two separated full meshs for cluster / corosync in a 3-node-cluster

mhert said:
By the way:

Can someone tell me which traffic goes through which connection on a cluster?

Throught which network goes traffic (oobe) of (builtin) backup / corosync / cluster (same as corosync?) / migration?

Is there an useful network diagramm of proxmox cluster with ceph?
Click to expand...
Anyone?
 
Hi together,
I'm planning the same network configuration, 1 mesh network (10G) for CEPH and 1 mesh network (1G) for corosync. I planned using the RSTP mesh for CEPH and corosync. As you guys (@admartinator @mhert) try using only frr, are there any disadvanteges in using RSTP ?
 
Hi, I have same problem, did you find some solution how to run two mesh networks using openfabric? Or did you use second as RSTP mesh?
 
As EllerholdAG said, I didn't get it to work using frr for both meshes. You can configure it, but then nothing works anymore...
 
i use a modified simple routed setup (one net for ceph and one for corosync) with some simple scripts.

https://pve.proxmox.com/wiki/Full_Mesh_Network_for_Ceph_Server#Routed_Setup_(Simple)

i use multiple ip addresses on each interface. (1st is "check IP" 2nd "service IP")
the first ip address of each interface is used to check if the direct link is availabe and active. (e. g. NIC1 A ==> NIC1 B)
==> yes: set routing service IP from A direct to Service IP B
if not check if the first ip of NIC 2 B is availabe via C ( NIC2 A ==> NIC 2 C ==> NIC 2B).
==> set routing service IP from A to Service IP B via C)
Service IP of NIC1 and NIC2 in each host is the same)

simple, working, routing changes is fast enough and in case of errors easier to handle for me then debugging frr, ospg etc.
 
  • Like
Reactions: Johannes S
You must log in or register to reply here.
Top Bottom