I have 1G and 2.5G networks for management and lan uplink with 10G for the cluster and 40G for ceph traffic.
Port layout is in the attached (bad) diagram.
Two questions:
Is this a sane way to do HA just using l2 bridges or is this asking for trouble? (STP, 1st node routing to the 3rd through the 2nd, etc) Is there a better way?
If not, what are the options to explore short of 10/40G network switches or learning OVS?
No plans for a 4th node in the HA cluster. I'll spare some details.
Secondly, can I simply trunk the 2.5G ports from the nodes to my switch(vlan aware) so VM guests don't lose connectivity during migrations without doing anything special?
I'm guessing yes, just figured I'd verify.
Thanks in advance.
Last edited: