On Mon, Nov 22, 2010 at 02:46:10PM +0100, Petr Šťastný wrote:
Hello,
what is the exact meaning of rfc1583compat switch in OSPF protocol a what is the appropriate difference in route table calculation?
There are two changes caused by rfc1583compat switch: 1) A subtle and tricky change in computing inter-area routes (or external routes from other areas, i forgot some details). This change is well described in RFC 2328 (RFC1583Compatibility flag). To get better understanding of that issue, i would also suggest to read relevant parts of RFC 1583 and RFC 2178 (where the problem with the old behavior is better explainted than in RFC 2328). Essentially, there was a bug in the OSPF algorithm in RFC 1583, which was (incompletely) fixed (in an incompatible way) in RFC 2328 and later (hope completely) fixed in RFC 2328. This flag forces to use original alg from RFC 1583. 2) There is also a BIRD-specific change in assigning LSA IDs for summary and external LSAs. With rfc1583compat, it is assigned in a way that is compatible with even older RFC 1247, but does not allow external (and summary) routes for nested subnets. See a comment in proto/ospf/topology.c in fibnode_to_lsaid() function for details. So, unless you really need some compatibility with some really archaic implementation, do not enable rfc1583compat. My advice is also not to use multiple OSPF areas at all. -- 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."