Help me understand why I cannot ping corosync interfaces

[crembz]

So I have setup my pve cluster with 3 VLANS

Primary interface is VLAN2
Corosync is VLAN3
Storage is VLAN4



VLAN 2 & 4 are running over an lacp 10g active link connected to a 10gb fiber switch
VLAN 3 is running over a 1gb link connected to the main switch

The fiber switch has an lacp 10g uplink to the main switch.

Now I was always under the impression that any traffic destined for the 10.3.0.x network would run over the VLAN3, which would mean that corosync would remain operational if the 10g links went down. However, if I disconnect the 10g links, I'm unable to ping 10.3.0.10 or access the management gui. With the links up I can ping and access the 10.3.0.10 interface.

That suggests to me one of two things, either the network traffic is traversing the 10g links/switch and then routed within the pve box to the 10.3.0.10 interface OR the box is using the default gateway (10.2.0.1) to respond to requests over the 10.3.0.10 interface.

Can someone please help me understand this a little better?

 

[crembz]

hrm ... the mystery deepens ... it seems I cannot receive traffic across vlans for some reason. I can ping each of the interfaces from any vlan but iperf against the pve host I will only receive traffic if the client is on the same vlan as the server. I'm at a loss with my limited network knowledge.