November 2020 - Bird-users

Bug: BFD confuses sessions with identical IPv6 link-local addresses
by Julian Brost 14 Jan '21

14 Jan '21

Hi, I just tried to enable BFD for some OSPF and BGP connections and ran into an issue which BFD sessions flapping up and down and Bird logging many messages like these: bfd1: Bad packet from fe80::2 - unknown session id (130079069) bfd1: Bad packet from fe80::2 - unknown session id (3102513000) bfd1: Bad packet from fe80::2 - unknown session id (3650438750) bfd1: Bad packet from fe80::2 - unknown session id (1597034259) This presumably happens due to this piece of code in bfd_rx_hook() in proto/bfd/packets.c: s = bfd_find_session_by_addr(p, sk->faddr); I use fe80::1/fe80::2 as addresses for most of my IPv6 peering connections and sk->faddr is just a 128 bit value, so it contains no interface identifier, which likely leads to Bird confusing multiple BFD sessions. Regards, Julian

3 5

[Bug] Static routes with recursive routes not going through filter when next hop changes
by Wang Shanker 14 Jan '21

14 Jan '21

Hi, all I recently spotted when filters used with recursive static routes can sometimes fail to work. The minimum case for reproducing this issue is as follows: ``` router id 192.168.0.1; table tab1; table tab2; protocol device { }; protocol static s1 { disabled; table tab1; route 2001:db8::1/128 via fe80::1%eth0; }; protocol static s2 { table tab2; igp table tab1; route 2001:db8::/32 recursive 2001:db8::1; import filter { print "Filter called"; print "dest=", dest; if dest = RTD_UNREACHABLE then { reject; } else { accept; } }; }; ``` Suppose in routing table `tab2`, we want the nexthop of 2001:db8::/32 follows 2001:db8::1 in table `tab1`, only when 2001:db8::1 is reachable. Hence, a filter rejecting unreachable routes is used. In `tab1`, a static protocol `s1` is used to emulate the routing changes there. In actual scenario, the nexthop of 2001:db8::1 may be decided by more complex routing protocols. When bird started, everything looks fine, both routing tables are empty: ``` # ./bird -c bird-test.conf -d -s bird.ctl & bird: Filter called bird: dest=(enum 34)3 bird: Started # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. ``` However, things begin to go wrong when I trigger a change in table `tab1`: ``` # ./birdc -s bird.ctl enable s1 BIRD 1.6.8 ready. bird: Enabling protocol s1 s1: enabled # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. 2001:db8::1/128 via fe80::1 on eth0 [s1 16:20:47] * (200) # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. ``` It seems that the changed route does not going through the filter and failed to be injected into table `tab2`. This also happens when the route is available at start and is withdrawn later. ``` # We edit the configuration and remove the "disabled" statement in protocol s1 # ./bird -c bird-test.conf -d -s bird.ctl & bird: Filter called bird: dest=(enum 34)0 bird: Started # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. 2001:db8::1/128 via fe80::1 on eth0 [s1 16:25:33] * (200) # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. 2001:db8::/32 via fe80::1 on eth0 [s2 16:25:33] * (200) # ./birdc -s bird.ctl disable s1 BIRD 1.6.8 ready. bird: Disabling protocol s1 s1: disabled # ./birdc -s bird.ctl sh route table tab1 BIRD 1.6.8 ready. # ./birdc -s bird.ctl sh route table tab2 BIRD 1.6.8 ready. 2001:db8::/32 unreachable [s2 16:25:32] * (200) ``` Expected behavior: I suggest that when there are any changes to the nexthop of recursive routes, they should be re-filtered. Cheers, Miao Wang

2 1

IPv6 link-local address with scope
by Wang Shanker 14 Jan '21

14 Jan '21

Hi, all I encountered a problem when trying to configure a static IPv6 route the nexthop of which is a link-local address. ``` protocol static { route 2001:db8:1::/48 via fe80::1234%foo; } ``` The above configuration worked well for routing traffic to 2001:db8:1::/48 to the router fe80::1234 at the interface named foo. However, if the name of the interface contains special characters, for example, `foo-bar`, bird will complain about the syntax. I dig into the source code and found out that in https://gitlab.nic.cz/labs/bird/-/blob/legacy/conf/confbase.Y#L162-165, `ipa_scope` is defined like `%[SYM]`, so in above scenario, the address shall be expressed as `fe80::1234%'foo-bar'`. I believe it could be better to have a record of this in user's doc. Cheers, Miao Wang

2 1

[Bug] Mixed BFD session for peers with same IPv6 link-local addresses on different interfaces
by Wang Shanker 14 Jan '21

14 Jan '21

Hi, all In bird, BFD sessions are distinguished only by their peers' address, so if multiple routing protocols request BFD session for the same neighbor, they can share and be controlled by one session. Normally it won't be a problem. However, it is not the case when it come to the IPv6 link-local addresses. Since fe80:: addresses are link-local, the same address on different interfaces refers to different nodes. So when requesting BFD sessions for neighbors with link-local address, the destination interface should also be considered. This issue can be reproduced by the following configuration with the latest commit in legacy branch, and a similar configuration with the latest commit in master branch: ``` router id 192.168.0.1; protocol bfd bfd1{ interface "eth*"; }; protocol device { }; protocol static { route 2001:db8::/32 multipath via fe80::1%eth0 bfd via fe80::1%eth1 bfd ; }; ``` As shown in `show bfd session`, only one BFD session is created: ``` # ./birdc sh bfd sess BIRD 1.6.8 ready. bfd1: IP address Interface State Since Interval Timeout fe80::1 eth0 Down 01:33:24 1.000 0.000 ``` Expected behavior: two BFD sessions are created with both routers on both interfaces. Cheers, Miao Wang

2 1

Template sans "channels"
by Clemens Schrimpe 14 Jan '21

14 Jan '21

Ahoj (again) :-) While in the process of cleaning up "ancient" configs (basically rewriting them from scratch) I came across an odd discovery: A template definition (for a kernel protocol) demands the specification of one (and only one) channel - which, in my case, defeats the whole purpose of having a kernel template in the first place. I attempted to have a template with all the "standard behaviour" I wanted (persist, metric, table #, ...) for both IPv6 and legacy-IPv4 kernel tables and then tried to apply it like so: template kernel KERNEL { persist; learn; metric 20; kernel table 10; } protocol kernel KERNEL6 from KERNEL { ipv6; } protocol kernel KERNEL4 from KERNEL { ipv4; } ... because I thought the templating mechanism was "just" some syntactic sugar, similar to an "include" statement?! As the kernel template demands a (single) channel statement, which could then, upon invocation, not be "negated" again, i.e., by a no ipv4; statement I can only have two templates - one pre-destined for IPv6 and another for IPv4. What is the intention behind this? I do understand, that -in the end- the actual protocol defined through a template must have one (and only one in the case of a "kernel" protocol) channel definition and in my example this requirement would be fulfilled. Thoughts? Hints? Thanks again - Clemens

2 2

Bird can't add static route.
by Mikhail Petrov 14 Jan '21

14 Jan '21

4 3

[PATCH] Nest: Read Babel metric as IGP metric
by James Lu 29 Dec '20

29 Dec '20

Hi all, When testing Babel with Bird 2.0.7, I noticed that unlike OSPF, Babel metrics are not read as an IGP metric for BGP purposes. I've attached a simple patch that adds this support - it seems to do the trick in my local testing. Best, James

4 8

Issue with unsigned integer LOCAL_PREF
by Alexandre Corso 04 Dec '20

04 Dec '20

Hello Bird team, Thank you for the great tool BIRD. I would like to open an issue (bird v2.0.7) about the local_pref. Today we hit an issue with one filter. if ! is_geo_region_local() then { bgp_local_pref = bgp_local_pref - 200; # If not defined, MED has a type of void instead of int if defined(bgp_med) then { bgp_med = bgp_med + 50; } else { bgp_med = 50; } } When some prefixes with the local_pref was below 200 (for example 100), the value overflow: localpref (unsigned integer 4^32) 100 - 200 = -100 = 4 ^ 32 - 100 = 4294967296 - 100 = 4294967196 We had the prefixes with this value 4294967196 X.X.X.X/32 unicast [ipv4_rr01 07:51:49.603 from 100.100.254.137] * (100/84285) [AS65030?] via 100.99.0.100 on eth1.666 Type: BGP univ BGP.origin: Incomplete BGP.as_path: 64619 65030 BGP.next_hop: y.y.y.y BGP.med: 50 BGP.local_pref: 4294967196 BGP.community: (35280,1040) (35280,2070) (35280,3110) BGP.originator_id: y.y.y.y BGP.cluster_list: 0.0.0.30 We had an exception to check the value of local_pref (on the filter) but can we have the something on the code, not to overflow the local_pref, for example SET to 0 if the value overflow? Our filter looks like this now if bgp_local_pref > 200 then bgp_local_pref = bgp_local_pref - 200; else { bgp_local_pref = 0; } Best Alexandre Corso alexandre@ac <mailto:alexandre@acorus.net>orso.fr

3 3

Inject ethernet-based defaults to OSPF
by daniele 04 Dec '20

04 Dec '20

Hello list, I???m looking for some advice from someone that has more experience. I'm running bird 1.6.6. I have two routers R1 and R2, each with a single internet uplink U1 and U2, running OSPF and sharing a LAN ethernet segment. Both U1 and U2 are dhcp-based ethernet segments themselves with a next-hop IP address that can change. I want R1 to be the primary LAN internet gateway and R2 the secondary. Should U1 fail, R1 has to route to R2 in order to reach internet through U2. How would you recommend to inject U1 and U2 default routes into OSPF? I cannot use the next-hop ip address as it changes along time, and an interface-based static route would not work. thanks a lot, d

3 8

BGP with ECMP and weight control
by Alexander Demenshin 02 Dec '20

02 Dec '20

Hi, I have two uplinks (different AS), and one is more stable/preferred, but when "merge paths" is set to on, all routes are generated with weight 1: 5.21.41.0/24 proto bird nexthop via 10.16.0.1 dev uplink1 weight 1 nexthop via 10.17.0.1 dev uplink2 weight 1 Is there any way to adjust weight for ECMP generated from BGP sessions? I could not find any option/attribute when it is related to BGP or kernel protocol, while it is possible for OSPF and static protocols (to a degree). Adjusting local preference or med (in BGP) does not really help - weight is unchanged, though of course I could force routes from specific AS to be preferred, but then I lose ECMP and if preferred session is dropped it takes time to restore connectivity. Thank you! Best regards, Alexander.

4 6