March 2002 - Bird-users - Mailing Lists

A bug??
by Damjan 10 Mar '02

10 Mar '02

Anyone knows what is this: 02-03-2002 14:48:14 <TRACE> SMSnet: State changed to feed 02-03-2002 14:48:14 <TRACE> SMSnet: Connected to table master 02-03-2002 14:48:14 <TRACE> SMSnet < added xxx.xx.xx.x/20 via xxx.xx.xx.x on eth2 02-03-2002 14:48:14 <TRACE> SMSnet: Sending UPDATE 02-03-2002 14:48:14 <BUG> sk_maybe_write: unknown socket type 134766456 it crashed my bird daemon. Bird is 1.0.5 -- Damjan Georgievski | Дамјан Георгиевски Skopje, Macedonia | Скопје, Македонија

2 1

Re: broadcast and multicast RIP on the same interface?
by noc＠clouddancer.com 08 Mar '02

08 Mar '02

In broadcast and multicast RIP on the same interface?, you wrote: > >Is there any reason why broadcast and multicast RIP can't be run on the >same interface? I haven't read the RFCs, but we'd only be using this in >a student lab setting with the multicast interface in mode "quiet", so I >think it should be okay. Well, don't waste too much time on RIP. It's pretty worthless in today's enviroment, subnetting/CIDR is too common. OSPF in bird can use this sort of debugging/experimenting. About every 5-6 hours, I lose all routing for 10 minutes. Usually I chain login to the remote machine (via all the intermediate machines), restart the protocol and get connectivity back quickly. The only clue I have is: Feb 20 12:47:00 rudder bird: Netlink: Network is unreachable Feb 20 12:47:38 rudder last message repeated 6 times I suspect it is a recent 2.4 kernel change somehow, or that bird does not have a robust 'maintain the link' technique. I ran bird before in a 2.0/2.2 kernel enviroment on a T1, no problems. The above outage occurs in 2.4 kernels on a 28k modem link (the desert does not have good telephone lines). Surely the decision to declare a dead link should depend on the link speed/bandwidth? OSPF/BGP4 cover all routing needs. When someone talks about IPv6, just smile and move away slowly, it will be years before it's prime time. Upstreams have simply raised the costs of retaining IPv4 addresses, which causes customers to divest their surplus and switch to NAT systems -- classic market actions, reduced supply causes smarter demand.

1 0

RIP-1 (RFC 1058) support, ipaddress/netmask prefixes questions
by birdie＠sigint.cs.purdue.edu 06 Mar '02

06 Mar '02

According to the docs, RIP version 1 is not supported, but looking at proto/rip/rip.c, it looks like it is. Which is correct? Also, the docs say that "ipaddress/netmask" (which I take to mean e.g. 10.0.1.0/255.255.255.0) is a valid form for prefixes, but filter/config.Y doesn't appear to allow it. (But hey, my yacc's pretty rusty.) BTW, kudos to the BIRD team. BIRD rains mucho guano on gated zebras.

2 1

broadcast and multicast RIP on the same interface?
by birdie＠sigint.cs.purdue.edu 06 Mar '02

06 Mar '02

Is there any reason why broadcast and multicast RIP can't be run on the same interface? I haven't read the RFCs, but we'd only be using this in a student lab setting with the multicast interface in mode "quiet", so I think it should be okay. Here's a patch to bird-1.0.5 to allow this: *** proto/rip/rip.c.orig Wed Aug 23 01:53:49 2000 --- proto/rip/rip.c Wed Mar 6 09:48:04 2002 *************** *** 770,789 **** if (c & IF_CHANGE_UP) { struct iface_patt *k = iface_patt_match(&P_CF->iface_list, iface); struct object_lock *lock; if (!k) return; /* We are not interested in this interface */ lock = olock_new( p->pool ); #ifndef IPV6 ! lock->addr = ipa_from_u32(0xe0000009); /* This is okay: we ! may actually use ! other address, but ! we do not want two ! rips at one time, ! anyway. */ #else ! ip_pton("FF02::9", &lock->addr); #endif lock->port = P_CF->port; lock->iface = iface; lock->hook = rip_real_if_add; --- 770,788 ---- if (c & IF_CHANGE_UP) { struct iface_patt *k = iface_patt_match(&P_CF->iface_list, iface); struct object_lock *lock; + struct rip_patt *PATT = (struct rip_patt *) k; if (!k) return; /* We are not interested in this interface */ lock = olock_new( p->pool ); + if (!(PATT->mode & IM_BROADCAST) && (iface->flags & IF_MULTICAST)) #ifndef IPV6 ! lock->addr = ipa_from_u32(0xe0000009); #else ! ip_pton("FF02::9", &lock->addr); #endif + else + lock->addr = iface->addr->brd; lock->port = P_CF->port; lock->iface = iface; lock->hook = rip_real_if_add;

2 1