Hello,
we have multiple vlan interfaces on some router, but they have the same IPv6 link local address as the parent ethernet interface. Is this normal or possible bug?
In log we have lot of entries: “route ospf error Discarding packet: locally originated”, because there is one vlan interafce and parent ethernet with ospfv3 and they have the same link local address.
thanks for suggestion
ll address is generated from mac address. Vlan mac address is the same as on physical interface.
Can the same instance of OSPFv3 work on multiple different VLAN interfaces of the same router (having the same link local addresses)?
My setup is a few routers, each with a single physical gigabit link connected to a managed switch, and everything done with multiple VLANs.
This simplifies the physical wiring, and I’m still far from saturating a gigabit link with combined traffic in all VLANs (the CPU might not handle more than gigabit anyway). It all works fine for IPv4, but now I’m starting to deploy IPv6 and have trouble with it.
R1, R2 = two x86 boxes running Vyatta 6.6, eBGP outside and iBGP between their loopback IPs over OSPF
R3. R4 = two RB1100AH running ROS 6.5, load balanced redundant PPPoE servers and VRRP for public services
R5, R6 = two RB2011L running ROS 6.5, NAT and routing between internal private networks
vlan42 = public /29 connecting R1, R2, R3, R4 with OSPF (R1, R2 with higher priority are DR, BDR)
vlan41 = public /27 connecting R3, R4 (sharing a VRRP default gateway IP), R5, R6 + a few servers on public IPs (+ global IPv6 /64 added with advertise option)
Now, I got a nice new public IPv6 range announced by BGP. OSPFv3 set up in vlan42 between R1, R2, R3, R4 - and IPv6 now works in vlan41 - so far so good.
Now, since I don’t need NAT for IPv6 - R5, R6 will only do simple routing and firewalling, but I still want redundancy and no manually configured static routes.
So, I’m trying to set up OSPFv3 in vlan41 between R3, R4, R5, R6. But, both R3 and R4 become two DRs after some waiting time, and there is no BDR.
If I disable OSPFv3 on R3. R4 vlan42 and restart (disable/enable) it on vlan41, it elects the DR and BDR correctly. So it works, but in only one VLAN at a time.
Any chance it would be fixed soon? Or do I have to add extra wires to different physical ports (different MAC, different link local addresses) to make it work?
Bump.
I’ve just checked and can confirm the issue still present in v6.7 - OSPFv3 doesn’t work properly on more than one VLAN interface (on the same physical ethernet port, sharing the same MAC and thus IPv6 link-local address) at a time.
It’s a known issue with some other router vendors as well, for example:
http://kb.juniper.net/InfoCenter/index?page=content&id=KB22439&actp=RSS
http://www.gossamer-threads.com/lists/nsp/ipv6/46226
Possible workaround - allow manually configured IPv6 link-local addresses (they could then be made different for each VLAN), but the second link above suggests that properly implemented OSPFv3 should look at (interface, address) pairs, not just address alone.
Any plans for MikroTik to fix this in v6 or is it too complex and only likely in v7?
I can confirm both RIPng and OSPFv3 have this problem in ROS 6.15.