Date: Tue, 28 Dec 1999 00:03:21 -0800 (PST) From: "Rodney W. Grimes" <freebsd@gndrsh.dnsmgr.net> To: joe@pavilion.net (Josef Karthauser) Cc: joe@FreeBSD.org (Josef Karthauser), cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: ports/net/zebra Makefile ports/net/zebra/files md5 Message-ID: <199912280803.AAA69605@gndrsh.dnsmgr.net> In-Reply-To: <19991223181337.L48740@florence.pavilion.net> from Josef Karthauser at "Dec 23, 1999 06:13:37 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
> On Thu, Dec 23, 1999 at 10:04:59AM -0800, Rodney W. Grimes wrote: > > > > It rather bothers me that zebra startup says it started ``ospfd'' > > when I have bgpd configured :-) > > > > 'Tis done. BTW Do you ospfd at all? I find that can't establish relationships > with machines on the local network unless there is a default route statically > set. The zebra guys seemed to imply that it was a bug in multicast routing > under FreeBSD. Any ideas? (Do I need MROUTING defined?) I have done some more digging on this problem tonight and the zebra folks probably said the right thing when the chalked this off as a bug in the multicast kernel code (I'll be looking at that now that I now know what the real problem is.) Here is what gave me the final push to go, yep this is a kernel bug (excert is from gated 3.5.11 src/krt_ipmulti_ttl0.c): /* The Deering multicast mods for the kernel require the existence of */ /* a default route (0.0.0.0), a default multicast route (224.0.0.0) or */ /* a specific multicast route to determine the default interface. */ /* This information is ignored since we explicitly set our interface */ /* when sending the packets, but lack of this information would prevent */ /* us from sending to the multicast address. To work around this we */ /* add our two multicast addresses to the routing table pointing at the */ /* loopback interface. If these were ever actually used we would get */ /* an error because the loopback interface does not support multicast, */ /* but since we do specify the interface before sending packets this */ /* should *never* happen. */ Now, if you want to use zebra w/ospfd, and don't want to have a default route the following 2 static routes in your zebra.conf file makes it work until I can find the real kernel bug: ! Static routes for broken Deering multicast code so that ospfd can send ! it's initial hello packets. ip route 224.0.0.5/32 127.0.0.1 ip route 224.0.0.6/32 127.0.0.1 Note that zebra does the wrong thing here too, it installs this as GATEWAY routes, not HOST routes, but hey, it don't matter in this case, we just need them in there to get things started, they aren't ever actually used to route a packet, _technically_ :-) -- Rod Grimes - KD7CAX @ CN85sl - (RWG25) rgrimes@gndrsh.dnsmgr.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199912280803.AAA69605>