On 2/5/22 18:59, Jay Vosburgh wrote: > Harry, > > I am attempting to reproduce the behavior you describe, but have been > unable to do so. Could you clarify some of the configuration specifics, > as follows: > > Starting with step 2, > > "2. On the host, create a bridge and vlan with two ports, each with the > chosen vlan as PVID and egress untagged. Assign those ports one each to > the guests as the interface, use e1000. Be sure to NOT autoconfigure the > host side of the bridge ports with any ip4 or ip6 address (including > fe80::), [...]" > > I have configured testbr0 with no addresses, i.e., "ip addr show": > > 15: testbr0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state > UP group default qlen 1000 > link/ether 8c:dc:d4:b3:cb:f1 brd ff:ff:ff:ff:ff:ff > > and added vnet1 and vnet3 (the interfaces that connect to the VMs), to > testbr0, and removing their respective fe80: addresses. I set the bridge > vlan behavior via > > bridge vlan add dev vnet1 vid 1234 pvid untagged > bridge vlan add dev vnet3 vid 1234 pvid untagged > bridge vlan del dev vnet1 vid 1 > bridge vlan del dev vnet3 vid 1 > > then added a separate Ethernet device to the testbr0, removed its fe80: > address, and set its bridge vlan as > > bridge vlan add dev eno50 vid 1234 > bridge vlan del dev eno50 vid 1 > > Adding addresses to the interfaces within the VMs results in ping between > the two functioning (even in the face of "ip neigh flush dev enp7s0"). > > At no time does "bridge mdb" affect the behavior (it lists no entries), > and it in unnecessary to add ff02:ffxx entries as you describe in step 6 > (I presume that your mention of "fe02::ff..." is a typo for "ff02"). > > I am testing with the Ubuntu 5.11.0-46 kernel, which differs slightly from > your 5.11.0-49. From which version did you upgrade (i.e., what was the > last known working version)? I'm preparing to test with 5.11.0-50 (I am > unable to locate -49), but would also like to know if the above > description matches your configuration. > > Thanks.
Also, I'd say if any of the interfaces *ever* had v6 addresses, then there would be entries in the fdb and mdb. But on these systems we have: net.ipv6.conf.all.autoconf = 0 before the interfaces are created. > -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to the bug report. https://bugs.launchpad.net/bugs/1959702 Title: Regression: ip6 ndp broken, host bridge doesn't add vlan guest entry to mdb To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1959702/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
