On Mon, Jun 27, 2022 at 03:47:34PM +0300, Roi Dayan wrote: > > could it be your bond0 configuration? > beside doing +bond0 into /sys/class/net/bonding_masters how do you > config bond0 and its slave? > > I don't have any static config file and use ip commands > as follows: > > ip link add name bond0 type bond mode $mode miimon 100 > ip link set dev $nic1 down > ip link set dev $nic2 down > ip link set dev $nic1 master bond0 > ip link set dev $nic2 master bond0 > ip link set dev bond0 up > ip link set dev $nic1 up > ip link set dev $nic2 up > > > then adding bond0 to ovs bridge, I see bond0 got ingress block 310 > but the slaves didn't get ingress. > > # tc qdisc show dev bond0 ingress > # ovs-vsctl show > eb7bb34b-bdf6-4b9b-88d8-22ffeaf47630 > ovs_version: "2.17.90" > # ovs-vsctl add-br ov1 > # ovs-vsctl add-port ov1 bond0 > # tc qdisc show dev bond0 ingress > qdisc ingress ffff: parent ffff:fff1 ingress_block 310 ---------------- > # tc qdisc show dev enp8s0f0 ingress > # tc qdisc show dev enp8s0f1 ingress > # ovs-vsctl show > eb7bb34b-bdf6-4b9b-88d8-22ffeaf47630 > Bridge ov1 > Port ov1 > Interface ov1 > type: internal > Port bond0 > Interface bond0 > ovs_version: "2.17.90" > # tc qdisc show dev bond0 ingress > qdisc ingress ffff: parent ffff:fff1 ingress_block 310 ---------------- > # tc qdisc show dev enp8s0f0 ingress > # tc qdisc show dev enp8s0f1 ingress > # > Hi, Roi. Thanks for having tested this patch.
I have reproduced your test case. It is because bond0 first opened with type == NULL, then auto_classified == true. When bond0 adds to ovs, bond0 opens with type == "system", and auto_classified will not be cleared. Will send v3. > > printing in the log showing master_netdev->auto_classified is 1. > you showed in your prev test its 1 as well . so how it got to be > 0 now for you to pass the if-statement? > Sorry, I copied the wrong log line. > _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
