ipv4 inside ipv6 for ospf
Hi. Does bird supports http://tools.ietf.org/html/draft-ietf-ospf-ipv4-embedded-ipv6-routing-14 ? If yes, how can i configure it? -- Vasiliy Tolstov,
On 23.6.2014 10:39, Vasiliy Tolstov wrote:
Hi. Does bird supports http://tools.ietf.org/html/draft-ietf-ospf-ipv4-embedded-ipv6-routing-14 ? If yes, how can i configure it?
Hi Vasiliy, unfortunately it does not. Ondrej
On Mon, Jun 23, 2014 at 10:46:16AM +0200, Ondrej Filip wrote:
On 23.6.2014 10:39, Vasiliy Tolstov wrote:
Hi. Does bird supports http://tools.ietf.org/html/draft-ietf-ospf-ipv4-embedded-ipv6-routing-14 ? If yes, how can i configure it?
Hi Vasiliy, unfortunately it does not.
Hi The draft (now published as RFC 6992) is just informational draft specifying how to use existing features of OSPFv3 to do that and BIRD already supports almost everything needed, so i would say it is more or less supported. You can just configure separate OSPFv3 instance (protocol block in BIRD), use 'instance' parameter for all interfaces in that OSPFv3 specifying chosen instance ID and use import and export filters to filter out other prefixes than the one used for IPv4-embedded IPv6 addresses. There are some minor details: 1) RFC 6992 assumes that routers supports RFC 5838, which BIRD does not, esp. it does not set AF-bit in Hello packets, which means that it may fail to establish adjacency with other implementations that implement RFC 5838. 2) BIRD does not support DN-bit (from RFC 4576, as requested by RFC 6992 section 9), but that should not cause problems. 3) BIRD would also push local IPv6 prefixes to IPv4-embedded-IPv6 topology, which are not needed here. This may be limited by 'stubnet ::/0 { hidden };' option for stub networks, but prefixes related to broadcast transit networks would still be added. Alternatively, you could just use regular OSPFv3 instance for propagating IPv4-embedded IPv6 prefixes together with regular IPv6 network topology. This would make the setup simpler, as you would not need separate OSPFv3 instances and IPv6 routing tables. This is a bit different approach than specified in RFC 6992, but it works in almost identically. -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."
2014-06-23 14:44 GMT+04:00 Ondrej Zajicek <santiago@crfreenet.org>:
Hi
The draft (now published as RFC 6992) is just informational draft specifying how to use existing features of OSPFv3 to do that and BIRD already supports almost everything needed, so i would say it is more or less supported.
You can just configure separate OSPFv3 instance (protocol block in BIRD), use 'instance' parameter for all interfaces in that OSPFv3 specifying chosen instance ID and use import and export filters to filter out other prefixes than the one used for IPv4-embedded IPv6 addresses.
There are some minor details:
1) RFC 6992 assumes that routers supports RFC 5838, which BIRD does not, esp. it does not set AF-bit in Hello packets, which means that it may fail to establish adjacency with other implementations that implement RFC 5838.
2) BIRD does not support DN-bit (from RFC 4576, as requested by RFC 6992 section 9), but that should not cause problems.
3) BIRD would also push local IPv6 prefixes to IPv4-embedded-IPv6 topology, which are not needed here. This may be limited by 'stubnet ::/0 { hidden };' option for stub networks, but prefixes related to broadcast transit networks would still be added.
Alternatively, you could just use regular OSPFv3 instance for propagating IPv4-embedded IPv6 prefixes together with regular IPv6 network topology. This would make the setup simpler, as you would not need separate OSPFv3 instances and IPv6 routing tables. This is a bit different approach than specified in RFC 6992, but it works in almost identically.
Nice. Thanks! Last question - does bird developers have plans to fix ecmp routing for ipv6? (Periodic scan routing table remove multipath routes) -- Vasiliy Tolstov,
On Mon, Jun 23, 2014 at 02:16:00PM +0400, Vasiliy Tolstov wrote:
Nice. Thanks! Last question - does bird developers have plans to fix ecmp routing for ipv6?
Definitely. The main question w.r.t. IPv6 ECMP in Linux is whether these fixes should be done to BIRD or to Linux kernel. -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."
participants (3)
-
Ondrej Filip -
Ondrej Zajicek -
Vasiliy Tolstov