Must bindnetaddr be configured for a separated cluster network?

[jimmymcheung]

I was trying out the proxmox functions, and I see that it is possible (and recommended) to have a separated network for cluster communication. Therefore I added link0 and link1 with a separated network (dual stack) and on a dedicated NIC port, and link2 and link3 the addresses with has gateway configured as fallback (which VM traffic will flow through), however, the cluster automatically chose the network address which has gateway configured. I looked up and find this article, that guiding to set up bindnetaddr. In the article it states:

This specifies the network address the corosync executive should bind to.

But I want to use another network (which carries VM traffic) as fallback, so at lease when de dedicated one fails, it has some reachability. Below is a more visual diagram explaining the settings.

eth2 (vmbr1)                               eth1 (vmbr0)
 |                                          |
 +- link1: 10.0.0.2/22 (v4, for cluster)    +- link2: 10.1.0.2/22 (v4, for VM traffic)
 +- link0: 2001::2/48  (v6)                 +- link3: 2002::2/48  (v6)
 +- gw:   NA                                +- gw:   10.1.0.1
 +- gwv6: NA                                +- gwv6: 2002::1

Currently link2 and link3 is chosen by corosync, no other nodes are added yet. From documentation, it was expected that lower digit of linkx should have higher priority, which is not happening. I’m configuring gateway on vmbr0 because I don’t want VM traffic go through cluster dedicated network.

PS. if I traceroute the router’s address of 10.0.0.2/22 network, it is the next-hop (no hop in between). So this link should not have problem.

 

[floh8]

use the docu:https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_redundancy

 

[jimmymcheung]

use the docu:https://pve.proxmox.com/pve-docs/pve-admin-guide.html#pvecm_redundancy

Yes, in the beginning I followed this article but it does not yield the result how it described:

If no priorities are configured manually (or two links have the same priority), links will be used in order of their number, with the lower number having higher priority.

Instead, it goes directly to a link with higher number even link with lower number is alive.

 

[floh8]

then the fall back start working....your first link has no connection

 

[jimmymcheung]

then the fall back start working....your first link has no connection

it does, I can ping to the router and any other devices on the network, it just due to the firewall that couldn't communicate to outside internet (e.g. 1.1.1.1)