May 2023 - Bird-users - Mailing Lists

Strange behaviour on filtering bgp_large_community ...
by Tim Weippert 20 May '23

20 May '23

Hi List, maybe i'm doing something wrong, but if i use some quick and dirty filter on the CLI all is working as expecting: show route where bgp_large_community ~ [(65000, 0, 1001)] show route where (65000, 0, 1001) ~ bgp_large_community both get the expected results, show all routes with the specific community attached. If i use it in a filter function to filter the routes out: function filter_large(int TestID) { if ( bgp_large_community ~ [(65000, 0, TestID)] ) then { return false; } return true; } i get "all" routes, if clause doesn't get catched. function filter_large_V2(int TestID) { if ( (65000, 0, TestID) ~ bgp_large_community ) then { return false; } return true; } with the V2 function i get what i want, filter out all routes with the community attached and show all the rest. I'm a little confused here, because the first version of the if clause is used for bgp_ext_community on several functions on my setup and working as expected, only for bgp_large_community i need to switch to if clause from version V2, and only in a function? Can someone confirm or explain this? regards, tim -- Tim Weippert http://weiti.org - weiti(a)weiti.org GPG Fingerprint - E704 7303 6FF0 8393 ADB1 398E 67F2 94AE 5995 7DD8

3 3

RFC 9234 (BGP roles) possible bug
by Mikhail Grishin 18 May '23

18 May '23

Hi, We ran BGP roles at 1000+ BGP sessions. About 0.5 - 1% of them affected by some issue. Probably all of them - Juniper with the old junos. Here is description: 1) Our device (Bird) sent BGP Open to the peers, with Capability: BGP Role Type: BGP Role (9) Length: 1 Unknown: 01 2) Some "broken?" peers respond with: Border Gateway Protocol - OPEN Message Marker: ffffffffffffffffffffffffffffffff Length: 29 Type: OPEN Message (1) Version: 4 My AS: xxxxx Hold Time: 90 BGP Identifier: 10.5.5.2 Optional Parameters Length: 0 Border Gateway Protocol - NOTIFICATION Message Marker: ffffffffffffffffffffffffffffffff Length: 21 Type: NOTIFICATION Message (3) Major error Code: Cease (6) Minor error Code (Cease): Connection Rejected (5) 3) After that, Bird genarates another packet: Border Gateway Protocol - NOTIFICATION Message Marker: ffffffffffffffffffffffffffffffff Length: 21 Type: NOTIFICATION Message (3) Major error Code: OPEN Message Error (2) Minor error Code (Open Message): Unknown (11) About stage 2) - peer (old junos) shouldn't generate such response due to === If a BGP speaker receives from its peer a capability that it does not itself support or recognize, it MUST ignore that capability. In particular, the Unsupported Capability NOTIFICATION message MUST NOT be generated and the BGP session MUST NOT be terminated in response to reception of a capability that is not supported by the local speaker. === At the same time, question to the stage 3) - why Bird gererate such message? + In the logs printed Error: Role mismatch (provider) Wbr, Mikhail.

4 5

[PATCH] BMP: Support for multiple protocol instances and bugfixes
by Zagorski, Michal 17 May '23

17 May '23

Hello, We've put some work in BMP to support multiple collectors and fix bugs found during testing. Please, review attached patch. As of now, known issue: * When BMP connects after BGP sessions established, the collector won't receive updates until peer down on BGP. -- Thanks, Michał Zagórski

1 0

unable to install leaked routes with krt_prefsrc set/how to unset an attribute?
by jo+bird＠swagspace.org 16 May '23

16 May '23

Hello. I am currently trying to get route leaking to work on Linux with bird. I have several VRFs and want to leak some of the routes from one VRF into an other. For this I have several routing tables in bird, one kernel protocol per table and pipe protocols to leak to routes between the tables. This works as intended. However I have a problem when exporting my leaked routes to the kernel. I get the log message May 16 15:28:24 host-1 bird[7349]: kernel4: 0.0.0.0/0: reinstalling May 16 15:28:24 host-1 bird[7349]: Netlink: Invalid argument May 16 15:28:24 host-1 bird[7349]: kernel4: 172.21.0.1/32: reinstalling May 16 15:28:24 host-1 bird[7349]: Netlink: Invalid argument Those 2 routes are not installed into the corresponding kernel routing table. These are exactly the routes that have the Kernel.prefsrc/krt_prefsrc attribute set. If I try to install these routes manually with all the attributes they have in the bird routing table I get the error > root@host-1:~# ip r add table 5 0.0.0.0/0 via 172.21.0.1 dev vlan-uplink-1 metric 1024 src 172.21.0.153 > Error: Invalid prefsrc address. This is because the IP address that is used here comes from an other VRF where it is available however it is not available in the VRF I am trying to install it. (I also don't need the prefsrc option to where the route is installed.) If I add the route to the Linux FIB without the prefsrc value or with an IP address that is available in the corresponding VRF, then it works out as intended. A workaround that I found is to set the krt_prefsrc attribute on export to something that is available in the target VRF. However it would be nicer to simply remove the attribute. So is it possible to remove an attribute from a route in bird? Greetings Jo

2 1

Re: Strange behaviour on filtering bgp_large_community ...
by Tim Weippert 15 May '23

15 May '23

Hi Maria, i think the confusion part is, that it works on CLI in a "symmetric" way but not in a function, that was my biggest questionmark :) regards, tim On Mon, May 15, 2023 at 01:21:46PM +0200, Maria Matejka wrote: > Hello! > > The ~ operator is not symmetric. We should probably better check its arguments or maybe even update the syntax. It basically means, when A ~ B, that A is in B. > > Sorry for that confusion. > Maria > > On 15 May 2023 12:59:11 CEST, Tim Weippert via Bird-users <bird-users(a)network.cz> wrote: > >Hi List, > > > >maybe i'm doing something wrong, but if i use some quick and dirty > >filter on the CLI all is working as expecting: > > > >show route where bgp_large_community ~ [(65000, 0, 1001)] > >show route where (65000, 0, 1001) ~ bgp_large_community > > > >both get the expected results, show all routes with the specific > >community attached. > > > >If i use it in a filter function to filter the routes out: > > > >function filter_large(int TestID) > >{ > > if ( bgp_large_community ~ [(65000, 0, TestID)] ) then { return false; } > > return true; > >} > > > >i get "all" routes, if clause doesn't get catched. > > > >function filter_large_V2(int TestID) > >{ > > if ( (65000, 0, TestID) ~ bgp_large_community ) then { return false; } > > return true; > >} > > > >with the V2 function i get what i want, filter out all routes with the > >community attached and show all the rest. > > > >I'm a little confused here, because the first version of the if clause is > >used for bgp_ext_community on several functions on my setup and working > >as expected, only for bgp_large_community i need to switch to if clause > >from version V2, and only in a function? > > > >Can someone confirm or explain this? > > > >regards, > >tim > > > >-- > >Tim Weippert > >http://weiti.org - weiti(a)weiti.org > >GPG Fingerprint - E704 7303 6FF0 8393 ADB1 398E 67F2 94AE 5995 7DD8 -- Tim Weippert http://weiti.org - weiti(a)weiti.org GPG Fingerprint - E704 7303 6FF0 8393 ADB1 398E 67F2 94AE 5995 7DD8

2 1

question regarding table lookups
by Andrew Azarov 15 May '23

15 May '23

Hello, I wanted to ask bird gurus whether there is a way to lookup record in a different table during filtering? We have several connections but also external policy generating daemon which optimizes routes. The route optimization goes into kernel table 2 (as opposed to master). After that we have a pipe which imports those with higher preference. Problem is when the route is withdrawn from bgp peers partly or completely (basically is not available any more) the route from peer table still gets imported. The result is - we have network routing to nowhere, while there might be alternative working routes. I tried to find any information on another table lookup inside filter but was unable to discover anything meaningful... Is there a way to either: a) lookup another table within import filter for master table and bump preference? b) during import from kernel table #2 bump preference in other tables? instead of using pipes c) modify preference inside tables manually without using kernel tables? ie with birdc? Thank you in advance Regards, Andrew

2 1

BIRD 3.0alpha2
by Ondrej Filip 15 May '23

15 May '23

Dear BIRD Users, we are releasing another alpha version of BIRD 3. We've found some annoying bugs and fixed some usability issues as well as several issues with BIRD crashing. We also added a simple work balancing algorithm between threads. MRT is still switched off, BMP is not merged; they will get into some future alpha version. Measured performance in BGP is approx. 6-10x better (with 20 threads) than version 2.13, depending on the actual configuration. Kernel synchronization is slow with a large number of routes; we will look into it soon. If you have some time for trying this version out, we would love to hear your feedback. I thank my colleagues and namely Maria for this release! Cheers Ondrej

5 13

Bird policy towards non RFC-compliant workarounds.
by Arnaud Houdelette-Langlois 12 May '23

12 May '23

Hi. I have an issue with the BGP implementation of a well-known hardware manufacturer. Said implementation send multiple Add-Path capabilities, one for each address family, instead of sending only one capability with both address families. Moreover, it expects it's BGP peer to behave identically. If bird manages to understand the badly formatted open message, the said BGP implementation fails and reject bird (well-formed) message. The manufacturer is informed, but has still to provide a working patch. In the meantime, I think it would be easy to implement a workaround as a bgp protocol option in bird. I'm ready to dive into the code to do it. But ... Is it bird policy to accept such workaround pull requests ? -- Arnaud Houdelette-Langlois Administrateur des infrastructures systèmes et réseaux Normandie Université +33 2 31 56 69 54

2 1

IGP(OSPF) learned route not used for iBGP?
by Nico Schottelius 12 May '23

12 May '23

Hello fellow bird users, a quick IGP/iBGP question: Why does apu-router1 that should learn the IPv6 default route from server138 complain about not accepting the route? May 12 17:13:18 apu-router1 daemon.err bird: server138_v6_v4: Invalid NEXT_HOP attribute - address 2a0a:e5c0:32:2::21 not directly reachable May 12 17:13:18 apu-router1 daemon.err bird: server138_v6_v4: Invalid route ::/0 withdrawn Checking that apu-router1 has the route: bird> show route 2a0a:e5c0:32:2::20/124 Table master6: 2a0a:e5c0:32:2::20/124 unicast [ospf6 17:14:01.612] * E2 (150/10/10000) [0.0.0.138] via fe80::3eec:efff:fed2:d0c2 on eth1.2 weight 1 via fe80::3eec:efff:fed2:d0c4 on eth2 weight 1 unicast [server137_v6_v4 17:13:19.524] (100) [i] via 2a0a:e5c0:0:b::89 on eth2 unicast [server138_v6_v4 17:13:18.394] (100) [i] via 2a0a:e5c0:0:b::8a on eth2 bird> The complete setup is as follows: apu-router1 -- server138 -- [wireguard] -- server120 - apu-router1 and server138 are ASN 199553 [iBGP) - server120 is 209898 [eBGP] As the wireguard interface seems to be unusable with OSPF (likely due to missing multicast support), we use direct to read the route from the interface, the snippet from server137: protocol direct { ipv6; interface "s*"; } And then we propagate it via OSPF: ipv6 { import all; # Forward the routes that we specifically picked from s* devices # that are not picked up by ospf export filter { if source = RTS_DEVICE then { accept; } reject; }; }; On server138: bird> show route ::/0 Table master6: ::/0 unicast [server121_p15_v6_v4 15:11:34.058] * (100) [AS209898i] via 2a0a:e5c0:32:2::21 on s138s121 bird> show route 2a0a:e5c0:32:2::20/124 Table master6: 2a0a:e5c0:32:2::20/124 unicast [direct1 15:11:12.662] * (240) dev s138s121 unicast [server137_v6_v4 15:11:36.524] (100) [i] via 2a0a:e5c0:0:b::89 on eth2 unicast [apu_router2_v6_v4 15:13:17.009] (100) [i] via 2a0a:e5c0:0:b::47 on eth2 unicast [apu_router1_v6_v4 15:14:01.616] (100) [i] via 2a0a:e5c0:0:b::46 on eth2 bird> What are we doing incorrectly so that the apu-router1 does not use 2a0a:e5c0:32:2::20/124 that it learned via OSPF? Best regards, Nico -- Sustainable and modern Infrastructures by ungleich.ch

4 5

OTC attribute
by Marek Küthe 11 May '23

11 May '23

Hello, I have a few questions about RFC 9234 or `local role role-name` in bird. Is the OTC attribute kept even if you don`t have a `local role`? If yes, how to filter out the OTC attribute? I would be very happy about an answer! Greetings Marek Küthe -- Marek Küthe m.k(a)mk16.de er/ihm he/him

2 1