June 2023 - Bird-users - Mailing Lists

issue with next hop using freebsd loopback?
by Benoit Chesneau 16 Jun '23

16 Jun '23

I first posted this issue on the freebsd ml but it's probably more an issue to post there. I have the loopbacks `1.1.1.1` and `1.1.1.2` of 2 nodes distributed using OSPF: ``` filter ospfexport { if (source = RTS_DEVICE) then accept; reject; } protocol ospf v2 { tick 2; ipv4 { import all; export filter ospfexport; }; area 0 { stub no; interface "vlan1", "vlan2" { type ptp; cost 5; }; interface "lo1" { stub; }; }; } ``` `1.1.1.1` is set on the loopback: ```# ifconfig lo1lo1: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384 options=680003<RXCSUM,TXCSUM,LINKSTATE,RXCSUM_IPV6,TXCSUM_IPV6> inet6 ::1 prefixlen 128 inet6 fe80::1%lo1 prefixlen 64 scopeid 0x9 inet 1.1.1.1 netmask 0xffffffff groups: lo nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL> # birdc show route for 1.1.1.1BIRD 2.13 ready. Table master4: 1.1.1.1/32 unicast [ospf1 15:56:43.054] ! I (150/0) [1.1.1.1] dev lo1 ``` The distribution works OK and I can ping each others: ``` # route -n get 1.1.1.2 route to: 1.1.1.2 destination: 1.1.1.2 gateway: 10.3.1.1 fib: 0 interface: vlan2212 flags: <UP,GATEWAY,HOST,DONE,PROTO1> recvpipe sendpipe ssthresh rtt,msec mtu weight expire 0 0 0 0 9000 1 0 ``` But the IP of the loopback can't be used as BGP next hop in the following configuration: ```protocol bgp rr_par_ipv4 { local 1.1.1.1 as 65001; neighbor 1.1.1.1.2 as 65001; default bgp_local_pref 40; rr client; ipv4 { import all; export all; next hop self; }; }; ``` The source IP used as next hop is not 1.1.1.1 but the IP `10.3.1.2` of vlan1 used for the connection between 2 nodes: ``` # ifconfig vlan1vlan2212: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 9000 options=680703<RXCSUM,TXCSUM,TSO4,TSO6,LRO,LINKSTATE,RXCSUM_IPV6,TXCSUM_IPV6> ether 50:65:f3:8b:98:71 inet 10.3.1.2 netmask 0xffffff00 broadcast 10.3.1.255 inet6 fe80::5265:f3ff:fe8b:9871%vlan2212 prefixlen 64 scopeid 0xd groups: vlan vlan: 20 vlanproto: 802.1q vlanpcp: 0 parent interface: mlxen0 media: Ethernet autoselect (40Gbase-CR4 <full-duplex,rxpause,txpause>) status: active nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL> ``` **Even if I force the next hop address (using `next hop address`) it still take the IP of the vlan. What's wrong?*** ``` $ bird --versionBIRD version 2.13 $ uname -nmr myhostname 13.2-RELEASE amd64 ``` Benoît

1 3

Bad Peer AS with 4B AS issue
by Łukasz Jarosz 15 Jun '23

15 Jun '23

Hello, I have an issue with connectivity towards provider's router (out of my control). I setup BGP session as usual: protocol bgp [redacted] { local [redacted] as [redacted]; neighbor [redacted] as [redacted]; ipv4 { table bgp; next hop self; import filter bgp_in; export filter bgp_out; }; } I think worth mentioning is that I have 4B AS whilst my peer has 2B AS. All I get is this error: BGP state: Idle Neighbor address: [redacted] Neighbor AS: [redacted] Error wait: 48.720/60 Last error: Received: Bad peer AS I checked pcap dumps and they seem reasonable. BIRD open is followed by peer OPEN with piggybacked NOTIFICATION message. Based on MAC address my peer is Juniper, but that is just my guess. I am running BIRD 2.0.2 (unfortunately straightforward upgrade is not exactly possible). Anyone bumped into similiar issue or can give me any pointers? Best regards, Łukasz Jarosz

2 3

BIRD 1 End Of Life Reminder
by Maria Matejka 15 Jun '23

15 Jun '23

Dear BIRD Users, we'd like to remind you that BIRD 1 is reaching its end of life on 31 Dec 2023. Based on our existing materials, we have prepared a short guide to upgrade to version 2 which you can also print out for your convenience (see the attachment). Please note that the upgrade may take some time; it's better to upgrade sooner than later. If you need some help, let us know. Maria and the BIRD Team -- Maria Matejka (she/her) | BIRD Team Leader | CZ.NIC, z.s.p.o.

1 0

Feature Request: Preference in bird
by Marek Küthe 15 Jun '23

15 Jun '23

Hello, I noticed that when you "mix" routing protocols, it can be useful to change the "preference", but unfortunately there is little documentation in bird about this and no way to view it. Is it possible to view the preference using "birdc"? Is it possible to include the default values (in https://gitlab.nic.cz/labs/bird/-/blob/v2.0.12/nest/route.h#L748-755) in the documentation? Greetings Marek Küthe -- Marek Küthe m.k(a)mk16.de er/ihm he/him

4 7

BIRD show bgp summary for cli
by Roman Bulakh 14 Jun '23

14 Jun '23

Hi, we use BIRD in our environment, and it would be very nice to have some command like `show bgp summary` in juniper. This command show information about peer IP, peer AS, last update, prefixes received/best/exported and last state. Example: bird> show bgp summary Peer AS Last state change Prfx rcvd/best/export State X.X.X.X asXXXX 2021-01-27 6786/6785/109000 Established Y.Y.Y.Y asYYYY 2020-12-16 0/0/0 Passive BGP Error: Hold timer expired The are patch created for BIRD 1.3.4 and 1.3.6, but it would be good to have this in upstream software, without rebuilding and patching. as I understand it, no functionality should be affected by these changes, this is only a visual update for cli. Old patches describes at pages: https://marc.info/?l=bird-users&m=132155489919052&w=2 http://trubka.network.cz/pipermail/bird-users/attachments/20120228/a1499be7… https://bird-users.network.narkive.com/Px6EH3yl/bird-1-3-6-show-bgp-summary…

4 3

Default table selection with 'show route' in bird 2.13
by Alexander Shikov 14 Jun '23

14 Jun '23

Hello! While testing bird 2.13 with a bunch of non-default tables, including separate tables for flow specifications, I've found the output of 'show route' without additional parameters a bit annoying: # birdc show route BIRD 2.13 ready. Table FLOW4: flow4 { dst 109.68.40.15/32; proto 17; dport 123; } [FLOWSPEC 2023-06-08] * (200) So I had to add additional "table all" parameter to get route from non-default IPv4 table: # birdc show route table all BIRD 2.13 ready. Table master4: Table master6: Table IP4: 109.68.40.15/32 unicast [RS0 2023-06-07 from 193.25.180.255] * (100) [AS25372i] via 193.25.180.17 on bxe0 [...] Table FLOW4: flow4 { dst 109.68.40.15/32; proto 17; dport 123; } [FLOWSPEC 2023-06-08] * (200) Second example: -- default behavior: # birdc show route 109.68.40.15/32 BIRD 2.13 ready. Network not found # birdc show route 109.68.40.15/32 table all BIRD 2.13 ready. Table IP4: 109.68.40.15/32 unicast [RS0 2023-06-07 from 193.25.180.255] * (100) [AS25372i] via 193.25.180.17 on bxe0 ^^ have to always add 'table all' -- changed behavior looks more logical to me: # birdc show route 109.68.40.15/32 BIRD 2.13 ready. Table IP4: 109.68.40.15/32 unicast [RS0 12:31:26.621 from 193.25.180.255] * (100) [AS25372i] via 193.25.180.17 on bge1 Thus in my opinion, when no tables specified, it's reasonable to select not just default tables but all tables with routes. Such behavior should be closer to other vendors, at least JunOS. Attached patch changes default table selection. Have a nice day! -- Alexander Shikov Technical Staff, Digital Telecom IX http://dtel-ix.net/

1 0

need help to use freebsd fib and bgp
by Benoit Chesneau 07 Jun '23

07 Jun '23

Hi all, What I am trying to achieve is to connect to remote peers using information found in fib1 routing table setup in freebsd and export / import routing information in this table using BGP. The issue I have right now is that this configuratin doesn't work. It can't connect to the remote A.B.C.0 ip and I get such errors in logs: ``` 2023-06-06 02:05:58.571 <TRACE> enkiv4: A.B.C.D/24: ignored 2023-06-06 02:05:58.571 <TRACE> enkiv4: A.B.C.D/32: [alien] created 2023-06-06 02:05:58.571 <TRACE> enkiv4.ipv4 > added [best] A.B.C.D/32 0L 4G unicast 2023-06-06 02:05:58.571 <TRACE> enkiv4.ipv4 < rejected by protocol A.B.C.D/32 0L 4G unicast ``` Is this pattern expected to work? I am wondering if it's not an issue in Freebsd 13.2p0 version. I tried different combination these days without success. Any help would be much appreciated :) The connection used for bgp arrived on a vlan 20 on nic2 which I setup on fib1, while the management interface is on another nic (bge0) and get its IP using DHCP: ``` bge0 ---> fib 0 (used for management) DHCP vlan20 -> fib 1 (connect to remote BGP peer) , static /31 IPV4 ``` Here is a snippet of network setup: ``` # ifconfig bge0 bge0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 options=c019b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4,VLAN_HWTSO,LINKSTATE> ether 94:18:82:7b:88:20 inet6 fe80::9618:82ff:fe7b:8820%bge0 prefixlen 64 scopeid 0x3 inet 192.168.88.249 netmask 0xffffff00 broadcast 192.168.88.255 media: Ethernet autoselect (1000baseT <full-duplex>) status: active nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL> # ifconfig vlan20 vlan2021: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 options=680703<RXCSUM,TXCSUM,TSO4,TSO6,LRO,LINKSTATE,RXCSUM_IPV6,TXCSUM_IPV6> ether 50:65:f3:8b:98:71 inet A.B.C.1 netmask 0xfffffffe broadcast 255.255.255.255 inet6 fe80::5265:f3ff:fe8b:9871%vlan2021 prefixlen 64 scopeid 0xb groups: vlan vlan: 20 vlanproto: 802.1q vlanpcp: 0 parent interface: mlxen0 fib: 1 media: Ethernet autoselect (40Gbase-CR4 <full-duplex,rxpause,txpause>) status: active nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL> $ sudo netstat -rn4 Routing tables Internet: Destination Gateway Flags Netif Expire A.B.C.0/31 link#11 U vlan2021 A.B.C.1 link#11 UHS lo0 127.0.0.1 link#7 UHS lo0 ``` A minimal setup (edited for anonymization) in Bird is the following ``` router id A.B.C.1; ipv4 table tabv4 sorted; protocol kernel enkiv4 { learn; kernel table 1; ipv4 { table tabv4; import all; export all; }; } protocol bgp transit_ipv4_1 { local A.B.C.1 as XXXX; neighbor A.B.C.0 as XXXX; default bgp_med 0; default bgp_local_pref 50; password "somepass"; ipv4 { table tabv4; import keep filtered; import filter ebgp_import; export filter ebgp_export; next hop self; }; }; ``` routes that will be adverised are setup as blackhole routes directly in FreeBSD. Should I setup them using the static protocol? Benoît

2 1

[PATCH] [RFC] Babel: Implement route daming with fixed delay
by Daniel Gröber 02 Jun '23

02 Jun '23

In order to prevent RTT based routing from causing persistent traffic oscillations we delay core rte announcement of each prefix by a configurable but metric invariant amount of time. Initial announcements and withdrawals will go through undelayed but a chnage in route for a prefix will cancel the previous announcement and reset the delay. --- This is just quickly thrown together to demonstrate the idea of how to replace metric smoothing with something more simplistic. Obviously the performance of using a linked list for keeping the scheduled announcements is suboptimal, our binary heap could be used instead. I'm not quite sure all the logic works out but let's discuss the concept before spending more time on it :) This could be extended to make the delay depend on metric difference if it turns out that's necessary I just skipped it for simplicity. proto/babel/babel.c | 79 +++++++++++++++++++++++++++++++++++++++++++-- proto/babel/babel.h | 11 +++++-- 2 files changed, 86 insertions(+), 4 deletions(-) diff --git a/proto/babel/babel.c b/proto/babel/babel.c index 2243fc89..d6d99a37 100644 --- a/proto/babel/babel.c +++ b/proto/babel/babel.c @@ -62,6 +62,7 @@ static inline int gt_mod64k(uint a, uint b) static void babel_expire_requests(struct babel_proto *p, struct babel_entry *e); static void babel_announce_rte(struct babel_proto *p, struct babel_entry *e, struct babel_route *r); +static void babel_sched_announce_rte(struct babel_proto *p, struct babel_entry *e, struct babel_route *r); static void babel_rte_update_unreachable(struct babel_proto *p, struct babel_entry *e, u8 announce); static void babel_send_route_request(struct babel_proto *p, struct babel_entry *e, struct babel_neighbor *n); static void babel_send_seqno_request(struct babel_proto *p, struct babel_entry *e, struct babel_seqno_request *sr, struct babel_neighbor *n); @@ -653,7 +654,7 @@ done: WALK_LIST2(r, n, nbr->routes, neigh_route) { r->metric = babel_compute_metric(nbr, r->advert_metric); - babel_announce_rte(p, r->e, r); + babel_sched_announce_rte(p, r->e, r); } } } @@ -772,6 +773,50 @@ babel_announce_rte(struct babel_proto *p, struct babel_entry *e, struct babel_ro } } +/* + * When route damping is enabled we have to delay updates to the kernel FIB. We + * do this by queueing some updates before actually announcing them to core. + */ +static void +babel_sched_announce_rte(struct babel_proto *p, struct babel_entry *e, struct babel_route *r) +{ + struct babel_config *cf = (void *) p->p.cf; + + if (!cf->route_damping || + !e->selected || + e->selected->metric < r->metric || + /*^ When the best route's metric is worse than the one we're about to + * announce it's safe to do so immediately even if route damping is on as + * it won't get exported to most other protocols. Note add-path BGP is + * still notified but we're not concerned about micromanaging the + * stability of exotic setups, only our local kernel FIB's impact. */ + r->metric == BABEL_INFINITY || !r->feasible) + /*^ This checks if announce_rte will do a withdrawal. */ + { + if (e->scheduled) { + rem_node(&e->sched); + e->scheduled = NULL; + } + + babel_announce_rte(p, r->e, r); + } + else + { + struct babel_iface *ifa = r->neigh->ifa; + + if (!e->scheduled) + add_tail(&ifa->sched_rtes, &e->sched); + e->scheduled = r; + + r->sched_time = current_time() + cf->route_damping; + + if (!ifa->want_sched) { + ifa->next_sched = r->sched_time; + ifa->want_sched = 1; + } + } +} + /* * Functions to send replies */ @@ -1371,7 +1416,7 @@ babel_handle_update(union babel_msg *m, struct babel_iface *ifa) e->updated = current_time(); } - babel_announce_rte(p, e, r); + babel_sched_announce_rte(p, e, r); } void @@ -1683,6 +1728,32 @@ babel_iface_timer(timer *t) p->triggered = 0; } + if (ifa->want_sched && now_ >= ifa->next_sched) { + TRACE(D_EVENTS, "Announcing scheduled rtes on %s", ifa->ifname); + + struct babel_entry *e, *ex; + + ifa->next_sched = TIME_INFINITY; + + u8 all_announced = 1; + WALK_LIST_DELSAFE(e, ex, ifa->sched_rtes) { + struct babel_route *r = e->scheduled; + ASSERT(r); + if (now_ >= r->sched_time) { + rem_node(&e->sched); + babel_announce_rte(p, e, r); + } else { + all_announced = 0; + if (r->sched_time < ifa->next_sched) + ifa->next_sched = r->sched_time; + } + } + + if (all_announced) { + ifa->want_sched = 0; + } + } + btime next_event = MIN(ifa->next_hello, ifa->next_regular); if (ifa->want_triggered) next_event = MIN(next_event, ifa->next_triggered); tm_set(ifa->timer, next_event); @@ -1862,6 +1933,8 @@ babel_add_iface(struct babel_proto *p, struct iface *new, struct babel_iface_con init_list(&ifa->msg_queue); ifa->send_event = ev_new_init(ifa->pool, babel_send_queue, ifa); + init_list(&ifa->sched_rtes); + struct object_lock *lock = olock_new(ifa->pool); lock->type = OBJLOCK_UDP; lock->addr = IP6_BABEL_ROUTERS; @@ -2523,6 +2596,8 @@ babel_init(struct proto_config *CF) P->rte_better = babel_rte_better; P->rte_igp_metric = babel_rte_igp_metric; + cf->route_damping = 15 S_; // HACK for testing + return P; } diff --git a/proto/babel/babel.h b/proto/babel/babel.h index 3868d7c4..186d4297 100644 --- a/proto/babel/babel.h +++ b/proto/babel/babel.h @@ -134,6 +134,7 @@ struct babel_config { list iface_list; /* List of iface configs (struct babel_iface_config) */ uint hold_time; /* Time to hold stale entries and unreachable routes */ u8 randomize_router_id; + uint route_damping; struct channel_config *ip4_channel; struct channel_config *ip6_channel; @@ -216,6 +217,7 @@ struct babel_iface { int tx_length; list neigh_list; /* List of neighbors seen on this iface (struct babel_neighbor) */ list msg_queue; + list sched_rtes; /* Scheduled rte announcments for route dampening */ u16 hello_seqno; /* To be increased on each hello */ @@ -226,7 +228,9 @@ struct babel_iface { btime next_hello; btime next_regular; btime next_triggered; - btime want_triggered; + u8 want_triggered; + btime next_sched; + u8 want_sched; timer *timer; event *send_event; @@ -292,6 +296,7 @@ struct babel_route { ip_addr next_hop; btime refresh_time; btime expires; + btime sched_time; }; struct babel_seqno_request { @@ -306,7 +311,9 @@ struct babel_seqno_request { }; struct babel_entry { - struct babel_route *selected; + node sched; /* for babel_iface.sched_rtes */ + + struct babel_route *selected, *scheduled; list routes; /* Routes for this prefix (struct babel_route) */ list sources; /* Source entries for this prefix (struct babel_source). */ -- 2.30.2

3 5

AdminDown is tearing down BGP session
by Sunnat Samadov 01 Jun '23

01 Jun '23

Hi all, We are using BIRD https://gitlab.nic.cz/labs/bird/-/tags/v2.0.12 version.

1 0