Hi, I have an annoying problem, of the "it used to work" kind of variety. The short version: I have two bridges on a linux box. Each bridge has an IP address. I connect two other boxes, one to each bridge, and set the appropriate bridge's IP address. However, my box won't route between the bridges. How can I get it to do so? The long version: (I know it is stupid to have just one interface per bridge, but I am using this example as it is the simplest example of the issue): I have three machines - A, B, and R: A[eth0] --- [tap0][br0] R [br1][eth1] ---- [eth0] B A is connected to tap0 of R, which is a member of br0 on R. B is connected to eth1 of R, which is a member of br1 on R. A is on 192.168.1.10/24. The default gateway is 192.168.1.1 (R:br0) B is on 192.168.2.20/24. The default gateway is 192.168.2.1 (R:br1) R:br0 is on 192.168.1.1/24 R:br1 is on 192.168.2.1/24 R is an Ubuntu 9.10 box with kernel 2.6.31 All firewalls are disabled (no rules, policy ACCEPT), and on R ip_forward = 1. If I sent a packet from A -> B (or B -> A), tcpdump shows it appearing on the first bridge, with a destination MAC address of the appropriate brX, but never on the second bridge. I think the kernel has received the packet onto the bridge, but either it has not accepted the packet (despite having a local destination MAC) or it has accepted the packet, but then decided not to forward it. It "used to work" and I have no idea what I have done to break it. I have modified the above setup to have '-j LOG' rules on INPUT, OUTPUT and FORWARD (but still policy ACCEPT), and the log files show *no* relevant entries at all, even though /proc/sys/net/bridge/bridge-nf-call-* are all '1'. I have tried STP on and off, rp_filter on and off. Does anyone have any idea what I am missing? Thanks, -- Jarrod Lowe _______________________________________________ Bridge mailing list Bridge@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/bridge
