November 2013 - Bird-users

Mail function
by Anibe Onuche 20 Nov '13

20 Nov '13

Hello I am suggesting that BIRD should have a mail function that alerts the administrator of critical errors in real time especially from it's syslog severs. Bird runs on a Linux platform and can be incorporated with some mailer daemon If a BGP section goes down.. I want Bird to alert me, If there is a change on my configuration.. I should be alerted etc. Would be grateful -- Anibe Onuche

5 6

AS numbers higher than 2000000000
by Milan Kocian 18 Nov '13

18 Nov '13

hi, I tried set up private 4b AS number (4200000000-4294967294). And I've discovered that my filters using int datatype doesn't work with these numbers. The reason is 'int' size (from -2000000000 to +2000000000). Does exist some other type for it (I didn't find anything in doc.) ? Best regards, -- Milan Kocian

3 3

More IPSEC routes for OSPF
by Iain Buchanan 18 Nov '13

18 Nov '13

Hi, I've a question about getting OpenSWAN and Bird routing working together. Hopefully someone here can point me in the right direction! There was an email from Michael Ludvig on the 8th July 2013 about exporting IPSEC routes for OSPF usage. The conclusion seemed to be that Bird can’t read the routes set up by OpenSWAN, and the only solution is to have a script running that exports the routes for Bird’s use into a separate table. I’m in pretty much the same position. I’ve tried Ondrej Zajicek’s suggestion of using transport mode IPSEC links, but this doesn’t seem to create visible routes (I’m using the netkey stack, which may be the issue). At the moment I’ve got GRE tunnels working on top of the IPSEC links, and if I enable debugging mode I can see instances of Bird communicating with one another over them (but not sending any of the OpenSWAN link information). I’d like to try and create routes that Bird can read, but I’m not certain how to extract the routing information nicely. (Part of the problem may be that the IPSEC links go through VPNs.) If I do an “ip xfrm policy” the output only refers to the gateways (i.e. where the NAT takes place), not the machines that I want to gain access to. I think what I need is a route saying “for remote_network/mask use the default route on this machine”. This could be written to a dedicated routing table which bird would read and send on. I’m guessing packets would then turn up and be processed by the xfrm rules and head on through the tunnel. It looks like the way to do this is to have a little service running that will regularly: (1) parse output from “ip route” to determine the default route (2) parse output from “ip address” to determine the set of local IP addresses (3) parse output from “ipsec auto status” looking for any of the “network diagram” lines that show the connectivity where one end or the other goes through a local IP address (4) update the routing table with calls to "ip route" (5) somehow prod Bird so that it reads the table I hope I’ve horribly over-complicated things and there’s an easier way to do this… does this sound like the way to go? Iain

6 11

Re: simple filters question
by Sergey Popovich 18 Nov '13

18 Nov '13

В письме от 9 ноября 2013 07:42:06 Вы написали: > Hi, > > I applied this code manually: > > + if ((v1.type == T_IP) && (v2.type == T_PREFIX_SET)) { > + v1.val.px.len = MAX_PREFIX_LENGTH; > + return trie_match_fprefix(v2.val.ti, &v1.val.px); > + } > > To filter.c > > However maybe the patch failed because I don't have the following: > > if (v2.type != T_SET) > return CMP_ERROR; > > > Maybe the patch that was sent was within a different branch than the main > 1.3.11 that's available for download? > Oh sorry, John, I patch really was made on top of current upstream tree, not 1.3.11. In attachment udpated patch version applied on top of 1.3.11 stable tree. Sorry for that. And yes, really according to documentation we could match only on set with elements of same tyme. On the other hand, an ip type may be easily represented as prefix with length of max AFI prefix length (32 - IPv4, 128 - IPv6, so I see no problem using this conversion of ip to prefix type. Moreover on IPv4 BIRD build there is implicit conversion of ip type to quad as on IPv4 they basically represent value with same format. Thanks for your note. -- SP5474-RIPE Sergey Popovich

3 2

Route table max up to 32K
by Bryan Duff 18 Nov '13

18 Nov '13

Any reason why the max route table is limited to 256 and not the Linux limit of 32767? Also any thoughts of adding dbus support or something similar for direct command injection (instead of rewriting the config and HUP'ing)? Thanks. -Bryan

4 3

simple filters question
by John Jensen 08 Nov '13

08 Nov '13

Hi, We're about to migrate our exchange to a /23 and we're likely going to do it over time. I have a simple filter in place to check and make sure that the NEXT_HOP is within our exchange's subnet. Originally it looked like this: function check_ixp_next_hop(ip nexthop) prefix ixpnet; { ixpnet = x.x.x.x/24; if ! (nexthop ~ ixpnet) then return false; return true; } And then within the inbound filter for each peer's ASN: if ! (check_ixp_next_hop(bgp_next_hop)) then reject; Since we're going to be caring about two distinct prefixes while we perform the migration, I changed the filter to look like this: function check_ixp_next_hop(ip nexthop) prefix set ixpnet; { ixpnet = [ x.x.x.x/24, y.y.y.y/23 ]; if ! (nexthop ~ ixpnet) then return false; return true; } However I'm getting the following errors in the log: 07-11-2013 12:33:00 <ERR> filters, line 61: ~ applied on unknown type pair 07-11-2013 12:33:00 <ERR> filters, line 61: ~ applied on unknown type pair 07-11-2013 12:33:04 <ERR> filters, line 61: ~ applied on unknown type pair Am I missing something? Thanks, -JJ

2 8

How bird configure multiple instances on ospf?
by 曾小小 06 Nov '13

06 Nov '13

I am now in use BIRD in the ospf routing protocol, such as cisco router I want to use multiple processes on the ospf. This how to achieve it ? Such as cisco router multi-process examples: r1 (config) # router ospf 100 r1 (config-router) # network 192.168.3.0 0.0.0.255 area 0 r1 (config-router) # network 10.0.5.0 0.0.0.255 area 0 r1 (config-router) # exit r1 (config) # router ospf 101 r1 (config-router) # network 10.0.7.0 0.0.0.255 area 1 My configuration is as follows , but I do not know if this configuration is not multi-process approach. filter export_OSPF { if (source = RTS_STATIC && net = 0.0.0.0 / 0) then { print "net accepted:", net; ospf_metric2 = 20; accept; } reject; } # # protocol ospf ospf_1 { table tab_ospf_1; router id x.x.x.x debug all; export filter export_OSPF; import all; preference 200; area 0.0.0.0 { interface "gre0" { hello 10; retransmit 5; cost 10; transmit delay 1; dead count 4; wait 40; type pointopoint; # type pointomultipoint; # type broadcast; priority 0; authentication cryptographic; password "abc"; } ; } ; } # # protocol ospf ospf_2 { table tab_ospf_2; router id x.x.x.x; # Debug all; export filter export_OSPF; import all; preference 200; area 0.0.0.0 { interface "gre1" { hello 10; retransmit 5; cost 10; transmit delay 1; dead count 4; wait 40; type pointopoint; # type broadcast; # type pointomultipoint; priority 0; authentication cryptographic; password "abc"; } ; } ; } Thank you..

1 0

Bird within netns (no OS routing table injection with BGP)
by Julien VAUBOURG 04 Nov '13

04 Nov '13

Hi all, I'm trying to use bird within netns with something like "ip netns exec foo bird". I can see received routes in "show route" but the injection in the OS routing table is not done. Strangely, static routes are correctly injected. With BGP routes I have no particulary debug messages, except "Netlink: Invalid argument" with bird6. Using Quagga, routes are correctly injected inside netns. Any idea about that problem? * Bird 1.3.7-1 * GNU/Linux (Debian Wheezy) Regards, ju.

2 2

FreeBSD, OSPF, and multicast
by Claude Marinier 01 Nov '13

01 Nov '13

Hi, I am building a WAN emulator using a FreeBSD server with DummyNet for WAN effects and BIRD for OSPF. At first, no multicast packets left the server; then I added a default route and they left by that interface only. How can I configure the server so BIRD sends HELLO packets out all the interfaces. Thank you. Details: FreeBSD WAMemu 9.2-RELEASE FreeBSD 9.2-RELEASE #0 r255898: Thu Sep 26 22:50:31 UTC 2013 root@bake.isc.freebsd.org:/usr/obj/usr/src/sys/GENERIC amd64 bird-1.3.11 P.S. I got it working by avoiding multicast with 'type nonbroadcast' and 'neighbors'. -- Claude Marinier

4 7