Issues with LACP and VM's

[sgt awesomesauce]

Hey all, I'm having trouble configuring a VM Bridge off of a LACP interface. I'll post my /etc/network/interfaces below, as I'm sure that will help. I'm using a TP-Link TL-SG1016DE operating a 2-port trunk, connecting to a PRO-1000 dual-port NIC. I'm not having any trouble with the interface or the actual LACP setup, as I can connect to the node's IP address (192.168.1.51) that's been assigned to the bond and get both SSH and the web-ui. My issue is that when I switch a KVM (don't have any OpenVZ setup, so I haven't tried) machine over to vmbr1 which I set up to link to bond0, I can't send or receive any packets. The minute I switch it back to vmbr0, which is set up to the onboard nic (I think it's a realtek, if memory serves), packets flow fine. It's not a huge issue, as I'm using this in a lab environment, but I'd like to be able to stop using the onboard nic for anything outside my management vlan.

# cat /etc/network/interfaces

# network interface settings
auto lo
iface lo inet loopback


iface eth0 inet manual


iface eth1 inet manual


iface eth2 inet manual


auto bond0
iface bond0 inet manual
        slaves eth1 eth2
        bond-miimon 100
        bond-mode 802.3ad
        bond-downdelay 200
        bond-updelay 200
        bond-lacp-rate 4


auto vmbr0
iface vmbr0 inet static
        address  192.168.1.50
        netmask  255.255.254.0
        bridge_ports eth0
        bridge_stp off
        bridge_fd 0


auto vmbr1
iface vmbr1 inet static
        address  192.168.1.51
        netmask  255.255.254.0
        gateway  192.168.0.1
        bridge_ports bond0
        bridge_stp off
        bridge_fd 0

I'm sure I'm missing something simple. I'm still very new to the non-basic (anything above subnetting, tbh) network topics.

Thanks,
Sgt

 

[Mr.Holmes]

Hello Sgt

My issue is that when I switch a KVM (don't have any OpenVZ setup, so I haven't tried) machine over to vmbr1 which I set up to link to bond0, I can't send or receive any packets. The minute I switch it back to vmbr0, which is set up to the onboard nic (I think it's a realtek, if memory serves), packets flow fine.

To which flow do you refer? Between an external client to the host or from/to the KVM?

Remarkable for me is that both bridges are membrs in the same subnet. So I suppose the bonded links are connectet to the same switch as eth0 is.


However, in order to have clear conditions I suggest to make tests using always categorically one bridge, both for VMs and Host. Remove from the other all address configuration.

If the problem persists check the packet flow with tcpdump.

Kind regards

Mr.Holmes

 

[sgt awesomesauce]

Mr. Holmes,
Flow from the KVM. My laptop can connect just fine regardless of if the KVM is connected or not.

I've reconfigured the Proxmox host to have eth0 disabled, and vmbr0 connect to bond0. I've deleted vmbr1 and moved all my vm's over to vmbr0.

Once I started my linux server (KVM), I could ping 192.168.1.51 (my PVE host) but nothing else on the network.

Also, my NFS share (which I use to host my non-critical VM's is not connecting. For some reason, I cannot ping the fileserver hosting it from the PVE host. It may be worth it to note that the fileserver is on the switch, configured as trunk 2. (trunk 1 being the PVE host in question)