[SOLVED] LACP not working

[foerkede]

I'm trying to set up LACP on my Proxmox using Openvswitch, but it's not working and I'm out of ideas and googling.
I have a Zyxel GS1900, enabled LACP and created a LAG-Group with the 2 ports connected to my server. In Proxmox i set up the network like this:

allow-vmbr0 bond0
iface bond0 inet manual
    ovs_bonds enp3s0 enp4s0
    ovs_type OVSBond
    ovs_bridge vmbr0
    ovs_options bond_mode=balance-tcp lacp=active other_config:lacp-time=fast

auto lo
iface lo inet loopback

iface enp2s0 inet manual

iface enp4s0 inet manual

allow-vmbr0 vlan100
iface vlan100 inet static
    address  10.0.100.5
    netmask  24
    gateway  10.0.100.1
    ovs_type OVSIntPort
    ovs_bridge vmbr0
    ovs_options tag=100

iface enp3s0 inet manual

allow-ovs vmbr0
iface vmbr0 inet manual
    ovs_type OVSBridge
    ovs_ports vlan100 bond0

The nagotiation seems to fail since both slaves are disabled:

---- bond0 ----
bond_mode: balance-tcp
bond may use recirculation: no, Recirc-ID : -1
bond-hash-basis: 0
updelay: 0 ms
downdelay: 0 ms
next rebalance: 2766 ms
lacp_status: negotiated
lacp_fallback_ab: false
active slave mac: 00:00:00:00:00:00(none)

slave enp3s0: disabled
  may_enable: false

slave enp4s0: disabled
  may_enable: false

---- bond0 ----
  status: active negotiated
  sys_id: 00:25:90:35:be:90
  sys_priority: 65534
  aggregation key: 2
  lacp_time: fast

slave: enp3s0: current attached
  port_id: 2
  port_priority: 65535
  may_enable: false

  actor sys_id: 00:25:90:35:be:90
  actor sys_priority: 65534
  actor port_id: 2
  actor port_priority: 65535
  actor key: 2
  actor state: activity timeout aggregation synchronized collecting distributing

  partner sys_id: 5c:e2:8c:f7:1b:d6
  partner sys_priority: 1
  partner port_id: 1
  partner port_priority: 1
  partner key: 1000
  partner state: activity aggregation defaulted

slave: enp4s0: current attached
  port_id: 3
  port_priority: 65535
  may_enable: false

  actor sys_id: 00:25:90:35:be:90
  actor sys_priority: 65534
  actor port_id: 3
  actor port_priority: 65535
  actor key: 2
  actor state: activity timeout aggregation synchronized collecting distributing

  partner sys_id: 5c:e2:8c:f7:1b:d6
  partner sys_priority: 1
  partner port_id: 3
  partner port_priority: 1
  partner key: 1000
  partner state: activity aggregation defaulted

If I run a tcpdump i see outgoing LACP-Packets, but I don't know why it doesn't negotiate properly. Since the logs on the switch are very limited (it doesn't output anything useful) I'm out of ideas what to do or how to investigate further.

 

[szhel]

What if change "allow-ovs vmbr0" to "auto vmbr0"?

 

[spirit]

What if change "allow-ovs vmbr0" to "auto vmbr0"?

you should keep "allow-ovs vmbr0", "auto vmbr0" give race condition on proxmox6 with ifup script.

do you have tried with linux bridge/bond to compare ? (like this you'll be sure it's coming from a miss configuration of the switch)

 

[foerkede]

you should keep "allow-ovs vmbr0", "auto vmbr0" give race condition on proxmox6 with ifup script.

do you have tried with linux bridge/bond to compare ? (like this you'll be sure it's coming from a miss configuration of the switch)

I did switch back to linux bridges, but that wasn't the problem. Today I did some more testing and finally found my error:

I already configured some VLANs on the switch and set "Accepted Frame Type" to "Tagged only" on the individual ports since the proxmox host should only send tagged frames in my config. What I didn't consider is that the LACP handshake stuff is blocked by that setting (because these frames don't get tagged). After I set the ports to also accept untagged Frames the bond works flawlessly.