- Bird-users - Mailing Lists

Add exceed merge path on limit
by Saklak, Marcin 13 Mar '25

13 Mar '25

Hi All, I’m wondering if it’s possible to add a log entry whenever the merged path limit is exceeded. For example, a route can have 64 next hops, but BIRD export to the kernel up to merge limit and skip rest of next hops. I’d like to be able to detect when this happens. Below is a potential solution for that: --- nest/rt-attr.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/nest/rt-attr.c b/nest/rt-attr.c index c8ef8e08..ef118cb0 100644 --- a/nest/rt-attr.c +++ b/nest/rt-attr.c @@ -323,6 +323,8 @@ nexthop_merge(struct nexthop *x, struct nexthop *y, int rx, int ry, int max, lin n = &((*n)->next); } *n = NULL; + if (max == -1) + log(L_ERR "Exceeded merge paths on limit!"); return root; } -- 2.34.1 If you have any suggestions or a better approach, please let me know! Best regards, Marcin Sakłak

2 2

OSPF type 5 LSA forward address change non zero to zero vice versa
by Bala Sajja 06 Mar '25

06 Mar '25

Hi Ondrej, If OSPF topology changes as part of interface addition/deletion to ospf area on router, that causes the forward address of type 5 lsa change to nonzero to zero or vice versa, Router should generate updated type 5 LSAs right with the updated forward address. Bird software seems not generating updated LSA's in this case. Regards, Bala.

1 0

Segment routing over IPv6 and OSPFv3
by Matyáš Kroupa 05 Mar '25

05 Mar '25

Hi, I recently decided that I would want to implement SRv6 extensions for OSPFv3 in Bird as my diploma theses. I study at the Faculty of Informatics, Masaryk University in Brno. As segment routing is a huge topic, I would start from the ground. That would mean implementing extended LSAs that use TLVs, then TLVs for Segment Routing with IPv6 dataplane. Since that would not be enough for a diploma thesis, I would also add definitions for Flex-algo. I already started with adding constants and data structures for extended LSAs. I read the contribution guide and worked on branch thread-next. I hope that my contribution will be welcomed and I am looking forward for hearing any comments from the maintainers. Best regards, Matyáš Kroupa

2 1

python library for generating bird configuration files
by Hendrik Wolff 03 Mar '25

03 Mar '25

Hey, I wrote a tiny library to generate bird configurations with python classes. It has no external dependencies and the repository includes some examples. It's in no way perfect or complete and I don't plan to extend it. But maybe it can serve as a base for your own needs. Extending it to support more protocols and options should be fairly easy. My personal favorite feature is that you can mark any Protocol class instance as a bird template by passing the is_template=True parameter. This template can be reused by passing the name of the template (e.g. "my_template") to the from_template="my_template" parameter of another Protocol class instance. https://github.com/hw0lff/bird-config-py Have fun! Cheers -- Hendrik

2 2

Interface specific RA DNSSL option behaviour
by Daniel Gröber 03 Mar '25

03 Mar '25

Hi BIRDies, I was playing with RA and DNSSL options today. I found the order of DNSSL options generated unhelpful. My config looks something like this: ``` protocol radv { dnssl { domain "global1.example"; domain "global2.example"; }; interface "enp2s0" { dnssl { domain "interface.example"; }; }; } ``` and I see the following RA options being generated (as shown by rdisc6): DNS search list : global1.example global2.example DNS search list lifetime: 1800 (0x00000708) seconds DNS search list : interface.example DNS search list lifetime: 1800 (0x00000708) seconds I tested with dhcpcd and rdnssd and saw problems with both: 1) dhcpcd outright doesn't support multiple DNSSL options so I only got the first one in /etc/resolv.conf (working on reporting) and 2) rdnssd supports this but the list in /etc/resolv.conf ends up being: search global1.example global2.example interface.example While rdnssd's behaviour seems in-line with RFC 8106 it's not what I wanted to achive overall. See section 6.3. "Synchronization between DNS Search List and Resolver Repository", https://datatracker.ietf.org/doc/html/rfc8106#section-6.3. > [...] it stores the DNSSL domain names (in order) [...] My goal was to have interface.example be the LAN domain (i.e the first in the search list) and the rest of the domains last in the search list. Admittedly this is a weird use-case and global dnssl was probably just intended for the simple case of all interfaces having the same search domain. If I had multiple LANs and corresponding domains I would not be able to share the list in the global dnssl option anyway, this really only works with LAN domains and network-wide search domain in the global list. I know I can fix this using the `dnssl local` switch (I tried and it works as expected), but perhaps a new option to give the order I want would be useful for others? Thanks, --Daniel

1 0

IPv4 routes with IPv6 nexthop in OSPFv3
by Lexi Winter 28 Feb '25

28 Feb '25

hello, it's me again, with my OSPF network. i was reading this blog post: https://ipng.ch/s/articles/2024/04/06/vpp-with-loopback-only-ospfv3-part-1/ which seems to suggest that Ondrej had done some work on advertising IPv4 prefixes into OSPFv3 with IPv6 nexthop in BIRD. however, it's not entirely clear to me how this works in practice, particularly since i don't use VPP. what i'd like to achieve is that each router that handles IPv4 packets has a single IPv4 address on lo0, and this somehow (magically) works with OSPFv3, so that i don't need to assign a /31 to every IPv4 OSPF peer. is this possible in BIRD 2.16.1, and if so, how would i go about setting it up? thanks, lexi.

2 1

advertise OSPF default route into a specific area?
by Lexi Winter 28 Feb '25

28 Feb '25

hello, i'm using BIRD 2.16.1 on FreeBSD 15.0. i have an OSPF ABR in area 0 and area 1. this router has a default route via a non-OSPF interface (kernel route, imported into BIRD using the 'kernel' protocol). i'd like to advertise this route into area 1, but *not* into area 0. how would i go about doing that? please feel free to let me know if this is a bad idea or if i'm using OSPF wrong :-) thanks, lexi.

3 3

Question about logging with 3.0.1
by Nigel Kukard 28 Feb '25

28 Feb '25

Hi there, I'm wondering if anyone has seen excessive filter logging with 3.0.0/3.0.1? Comparing log output from tests done on v2 vs v3, I'm seeing 2G+ log files generated in 30s on v3, when in v2 the log files were only a few hundred Kb in size at most after some time. I have router R1 and R2 connected via BGP. I have a couple of filters which log if a prefix has been filtered or accepted, I've included my log configuration and filter below. Here is the log config I'm using... log "/var/log/bird.log" { info, warning, fatal }; debug protocols { states, routes, filters, interfaces, events }; timeformat base iso long; timeformat log iso long; timeformat protocol iso long; timeformat route iso long; Here is the test filter I created to test... define DEFAULT_ROUTE_V4 = 0.0.0.0/0; define DEFAULT_ROUTE_V6 = ::/0; function is_default(string filter_name) -> bool { if ((net.type = NET_IP4 && net = DEFAULT_ROUTE_V4) || (net.type = NET_IP6 && net = DEFAULT_ROUTE_V6)) then return true; return false; } function bgp_accept_bgp(string filter_name) -> bool { if (is_default()) then return false; print filter_name, " [bgp_accept_bgp] Accepting BGP route ", net, " from ", proto; accept; } filter f_bgp_master_export string filter_name; { filter_name = "f_bgp_master_export"; bgp_accept_bgp(filter_name); reject; }; Here is the odd behavior I"m observing in v3... 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.111.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.112.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.113.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.201.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.131.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.132.0.0/24 from bgp4_AS65000_r1 repeating very many times per second... 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.111.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.112.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.113.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.201.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.131.0.0/24 from bgp4_AS65000_r1 2025-02-24 15:21:04 [0002] <INFO> f_bgp_master_export [bgp_accept_bgp] Accepting BGP route 100.132.0.0/24 from bgp4_AS65000_r1 In about 30s my log file grew to 2.1GB before I stopped the process. Perhaps I'm missing something in the migration guide? any help would be greatly appreciated. Kind Regards Nigel

2 4

[PATCH] Fix source-specific routing in Bird 3
by Toke Høiland-Jørgensen 26 Feb '25

26 Feb '25

Commit 69d1ffde4c72 ("Split route data structure to storage (ro) / manipulation (rw) structures.") changed rte->net from a pointer to a 'struct network' to a 'struct net_addr', but kept the address-of (&) operator before casting to 'net_addr_ip6_sadr *' when sending a source-specific route to the kernel. Because 'struct network' had an embedded struct member (struct fib_node), the address-of was needed to get back to a pointer to the data, but with the change in the commit mentioned above, e->net is now a straight pointer to the address. The bug meant that the source prefixes passed to the kernel were essentially garbage, leading to routes in the kernel like: default from b74:9e05:0:1:d8cf:c000::/86 via fe80::1 dev eth0 proto bird metric 32 pref medium Fix this by getting rid of the address-of operator. Fixes: 69d1ffde4c72 ("Split route data structure to storage (ro) / manipulation (rw) structures.") Signed-off-by: Toke Høiland-Jørgensen <toke(a)toke.dk> --- sysdep/linux/netlink.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sysdep/linux/netlink.c b/sysdep/linux/netlink.c index 1703a67595d5..6cc0142c3e00 100644 --- a/sysdep/linux/netlink.c +++ b/sysdep/linux/netlink.c @@ -1494,7 +1494,7 @@ nl_send_route(struct krt_proto *p, const rte *e, int op) /* Add source address for IPv6 SADR routes */ if (e->net->type == NET_IP6_SADR) { - net_addr_ip6_sadr *a = (void *) &e->net; + net_addr_ip6_sadr *a = (void *) e->net; nl_add_attr_ip6(&r->h, rsize, RTA_SRC, a->src_prefix); r->r.rtm_src_len = a->src_pxlen; } -- 2.48.1

2 2

Bird v3 failed assert
by Maiyun Zhang 25 Feb '25

25 Feb '25

Hello folks, I was checking the log today and saw a failed assertion in bird. bird[43579]: Assertion '!bmap_test(&c->export_rejected_map, old->id)' failed at nest/rt-table.c:1571 After this, the daemon dies and gets restarted by systemd. The frequency of this fault ranges from once every few minutes to a few hours depending on BGP traffic. I am running the latest build on branch stable-v3 (d13c9d3), built with gcc 13. This error happens on all three of my routers running the same revision. All of them are aarch64 linux machines. Two of them have eBGP peers and one is iBGP only (and babel). Judging from the log lines, there is not a pattern before which this happens. The set up is a little complex for me, so I’m sorry for not having a ready-to-go MWE. Please let me know if I can help debugging it. Thank you all for the great work in Bird! Best, Maiyun Zhang (he/him)

2 1