January 2019 - Bird-users

reorder BGP plink preference
by Eugene M. Zheganin 27 Jan '19

27 Jan '19

Hello, I have a bird router serving multiple BGP sessions with my AS uplinks, and now I need to create a more complicated thing, - I have several FIBs and I need to reorder the BGP link preference for a given FIB. Say, I have 3 uplinks, A/B/C, A is the most preferable one for default FIB, but I need B to be the most preferable one for FIB 1. I have some basic experience with multiple routing tables in bird, I'm successfully using two at the moment, but I just have no idea on how to do this. I mean I know how to make the B upstream the only channel for FIB 1 (without redundancy), but I have no idea on how to reorder the preference in order to keep the redundancy. Is there such a way ? Thanks. Eugene.

2 2

A few suggestions for docs improvement
by Igor Podlesny 26 Jan '19

26 Jan '19

First of all, protocols description are better kept in separate HTML pages otherwise it's hard to find by some keywords that several protocols have. It's really inconvenient to loose focus when you search something related to OSPF and Find scrolls page to completely different protocol. Second: why is font so tiny? -- End of message. Next message?

3 2

OSPF transmit delay is not documented?
by Igor Podlesny 26 Jan '19

26 Jan '19

Whereas it's given in some default Bird configs, I couldn't find it in manual. -- End of message. Next message?

2 1

Bird OSPF bug? (Mutliple areas, rfc1583compat off, cost ignored)
by Igor Podlesny 26 Jan '19

26 Jan '19

I have test lab -- two OSPF routers, Bird 1.6.4: A, B. 2 areas defined: 0.0.0.0, 0.0.0.1. Export "all". When rfc1583compat is Off I observe default route gets propagated via area 0.0.0.1 in despite its interfaces have higher costs than area 0.0.0.0's have. Turning rfc1583compat On changes it to expected behaviour: interfaces costs are respected again. Bug or am I missing some piece of OSPF knowledge here? -- End of message. Next message?

2 3

OSPF, BSD: every additional IP on an interface gives another neighbour
by Igor Podlesny 26 Jan '19

26 Jan '19

The docs say: "... secondary switch On BSD systems, older versions of BIRD supported OSPFv2 only for the primary IP address of an interface, other IP ranges on the interface were handled as stub networks. Since v1.4.1, regular operation on secondary IP addresses is supported, but disabled by default for compatibility. This option allows to enable it. The option is a transitional measure, will be removed in the next major release as the behavior will be changed. On Linux systems, the option is irrelevant, as operation on non-primary addresses is already the regular behavior. ..." Let's focus on: "... Since v1.4.1, regular operation on secondary IP addresses is supported, but disabled by default for compatibility. This option allows to enable it. ..." My observations show that what actually happens when whether this option is On or it's Off is not anything regular. Instead for every additional IP there's "virtually" additional OSPF neighbour instance. bird> show ospf ospf1: RFC1583 compatibility: disabled Stub router: No RT scheduler tick: 1 Number of areas: 1 Number of LSAs in DB: 23 Area: 0.0.0.0 (0) [BACKBONE] Stub: No NSSA: No Transit: No Number of interfaces: 3 Number of neighbors: 3 Number of adjacent neighbors: 3 -- every IP-address on interface actually handled as separate interface with that IP, that's why this weirdness with neighbour count is seen. O_o Is it so on Linux as well, or BSD implementation specific? -- End of message. Next message?

2 1

i-bgp neighbor not coming up
by Shahan Agha 25 Jan '19

25 Jan '19

Hi All, I am facing something very strange from almost a week now. I have multiple i-bgp neighbors which are working just fine. # birdc show protocols bgp_r4 BGP master up 2019-01-21 Established bgp_r3 BGP master up 2019-01-21 Established *bgp_r2 BGP master start 15:04:30 Active Socket: Connection refused* There as in issue with only one of them. It did work initially but for some reason it is now down since a few days. I have tried everything in my knowledge but have failed to bring it to work again. Below are the configurations: # R-1 # protocol bgp bgp_r1 { # disabled; debug { states, routes, filters, interfaces, events }; router id 10.10.2.1; local as 65001; neighbor 10.10.2.2 as 65001; next hop self; direct; } # R-2 # protocol bgp bgp_r2 { # disabled; debug { states, routes, filters, interfaces, events }; router id 10.10.2.2; local as 65001; neighbor 10.10.2.1 as 65001; next hop self; direct; } Any idea what can be the possible reason behind this? Thanks, Shahan

2 2

BGP route back
by Roman Romanyak 22 Jan '19

22 Jan '19

Hello Bird users, Does anyone know if there is a way to announce a BGP route back to the router it was received from. I need to do this for the following scenario. Lets say there are two routers with 2 transit links on each with full view tables and Bird server as a route-reflector. There is a need to force traffic to a specific destination via one of the ISPs, so I match the route in the import filter and set a local pref on it. But that will only make the route server announce the route with a high local pref to a second router, where the desired transit isn't directly connected. So I get the desired effect on the second router, it will learn and install a route with high local pref. The first router where the desired transit link is connected by default selects another transit link because of the shorter as-path. Here is the import filter snippet (x.y.z.0/24 is a dest route, as-path 1234 is a directly connected ISP on router-1: if source = RTS_BGP && net = x.y.z.0/24 && bgp_path.first = 1234 then { bgp_local_pref = 150; accept; } I think that bird doesn't do that because the protocol matches on the peer and on the route. Thanks!

3 8

Upcoming master branch reset in BIRD repository
by Maria Jan Matejka 22 Jan '19

22 Jan '19

Dear users, we are announcing here that we are going to rename branches in BIRD repository. 1st step (already done): a new branch >> legacy << is now the main branch for 1.6.x development. 2nd step (to be done between Jan 22 and Jan 24): the >> master << branch will be reset to what is currently >> int-new << and will be the main branch for 2.x development 3rd step (to be done until end of January): the >> int-new << branch will be deleted Please update your local repositories. Thank you for your understanding. Maria BIRD developer

2 1

add "src" attribute to kernel route
by Volodymyr Litovka 22 Jan '19

22 Jan '19

Hi colleagues, in Linux it is possible to specify source IP address for all packets with undefined this field, like: root@n1:~# ip route 10.0.0.0/24 dev eth1 proto kernel scope link src 10.0.0.1 10.1.0.2 via 10.0.0.2 dev eth1 src 10.1.0.1 using the command root@n1:~# ip route add 10.1.0.2/32 src 10.1.0.1 nexthop via 10.0.0.2 dev eth1 (note "src" parameter) This configuration instructs kernel to assign specified source address for all outgoing packets which travel to specified destination, if source address in the packet is unspecified. This quite important in order to provide reliability for applications which can't specify source IP address. in general, for such apps Linux chooses IP address of outgoing interface and, when this interface gone, all sessions go away as well since they're binded to interface's address. If I will be able to assign Loopback address, I will always have sessions active regardless of physical interfaces' state. Whether it's possible to configure Bird (We're using 1.6 on Ubuntu 18) in same way - when installing routes into the kernel, add "src" parameter like in the example above? Thank you. -- Volodymyr Litovka "Vision without Execution is Hallucination." -- Thomas Edison

2 1

RE: Bird peer disabled via control socket is automatically getting enabled
by Mohandass, Roobesh 22 Jan '19

22 Jan '19

Resending this email again, as I am not subscribed to this mailing list bird-users(a)network.cz<mailto:bird-users@network.cz> From: Mohandass, Roobesh Sent: Thursday, January 17, 2019 12:36 PM To: 'bird(a)network.cz' <bird(a)network.cz> Subject: Bird peer disabled via control socket is automatically getting enabled Hi Team, We are using BIRD version 1.6.3 in ubuntu 18.04 LTS kernel 4.15.0-42-generic, use case below. I have disabled few bgp peer using 'birdc disable <peername>' and after an hour or so, I just seen the peers are enabled back automatically, but I have not run 'birdc enable <peername>'. So thinking, was there anything that is reloading the bird configure at specific interval automatically or enabling the disabled peer automatically ? I have not configured anything related to below ones to be sure, configure [soft] ["config file"] [timeout [num]] Side Note: we are aware of commenting the peers which we are thinking to disable and then reload the config to make it always in disabled state. But just would like to check, why when we disable the peer from bird control socket , we observe this behavior. Thanks for your attention on this message. -Roobesh G M

2 4