March 2017 - Bird-users - Mailing Lists

State of OSPF/ECMP Multipath on FreeBSD
by chrono 24 Mar '17

24 Mar '17

Ahoy all, I've been asking on IRC and tried to find a definitive answer in the ML archives, but I'm still not sure what the current state is. Our Setup: BIRD (1.4.0 FreeBSD 10.x) <- OSPF/MPLS -> BIRD (1.5.0 FreeBSD 11.x) There were some msgs back in 2014 about merging ECMP support for FreeBSD but when I enable it, no more traffic is going through. I can see the Multipath setup in birdc though. Some users on IRC suggested it would only work on Linux but in this case I'm not free to decide which OS to pick and bound to FreeBSD. Did I miss anything or is there really no working ECMP support for bird on FreeBSD yet? If not, is there some sort of bounty possibility? I could check with my company if they're open to it (since we really need it) - if that could help bring that feature forward. -- chrono

2 1

Bird BGP session sending default route
by Michael McConnell 24 Mar '17

24 Mar '17

Hello all, Currently I receive a full route table from our uplinks on our primary routers using BGP, however I am trying to have a downstream bird router / BGP session where I only pass this downstream bird session a default route / partial table. I’ve experimented with the multihop parameters as that is the way I would effectively force a eBGP session in Cisco but not having luck transferring that to bird. Any suggestions how I can create a eBGP session so that I basically only receive a default route instead of the full route table? Thanks so much for your help, I really appreciate it! Mike -- Michael McConnell WINK Streaming; email: michael(a)winkstreaming.com <mailto:michael@winkstreaming.com> phone: +1 312 281-5433 x 7400 cell: +506 8706-2389 skype: wink-michael web: http://winkstreaming.com <http://winkstreaming.com/>

1 0

Filtering question
by Alarig Le Lay 20 Mar '17

20 Mar '17

Hi, I have this configuration: function bgp_filter_customer_in (prefix customer_prefix) { if ! (net ~ customer_prefix) then return false; else return true; } template bgp PEERS { local as myasn; import keep filtered; import all; export all; import limit 100 action restart; rs client; source address 2001:7f8:b1::1; } filter bgp_filter_grifon_nominoe_in { if (bgp_filter_customer_in([ 2a00:5884::/32, 2001:678:3cc::/48 ])) then { accept; } else { reject "Prefix filtered for Grifon"; } } protocol bgp bgp_grifon_nominoe from PEERS { description "Grifon nominoe"; neighbor 2001:7f8:b1::a as 204092; import filter bgp_filter_grifon_nominoe_in; } But the two announced prefixes are filtered: bird> show route filtered protocol bgp_grifon_nominoe 2a00:5884::/32 via 2001:7f8:b1::a on eth2 [bgp_grifon_nominoe 16:05:53] (100) [AS204092i] 2001:678:3cc::/48 via 2001:7f8:b1::a on eth2 [bgp_grifon_nominoe 16:05:53] (100) [AS206155i] If I just pass one prefix in the filter, this prefix is accepted, so I don’t understand why they are filtered with this configuration. Plus, I used the same syntax for the IPv4 prefixes and it works. I’m using bird 1.6.3. Thanks for your advices, -- alarig

2 2

[PATCH 00/13] Some more cleanup for configure.ac and aclocal.m4
by Ruben Kerkhof 18 Mar '17

18 Mar '17

Ruben Kerkhof (13): aclocal.m4: remove unused macro aclocal.m4: use normal macro for function use fixed width integers types from <stdint.h> Remove check for struct ip_mreqn aclocal.m4: consistently quote and indent configure.ac: fix indentation configure.ac: use AS_HELP_STRING for help text Consistently quote all macro arguments Remove dead code configure.ac: reorganize a bit Simplify check for alloca.h Simplify check for syslog.h No need to look for clock_gettime in libc aclocal.m4 | 321 ++++++++++++++++---------------- configure.ac | 502 +++++++++++++++++++++++++++++---------------------- sysdep/config.h | 22 ++- sysdep/linux/sysio.h | 10 - sysdep/unix/log.c | 8 +- 5 files changed, 459 insertions(+), 404 deletions(-) -- 2.12.0

1 11

[PATCH 02/13] aclocal.m4: use normal macro for function
by Ruben Kerkhof 18 Mar '17

18 Mar '17

--- aclocal.m4 | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/aclocal.m4 b/aclocal.m4 index a13b71f3..36749c58 100644 --- a/aclocal.m4 +++ b/aclocal.m4 @@ -134,7 +134,7 @@ AC_DEFUN([BIRD_ADD_GCC_OPTION], # BIRD_CHECK_PROG_FLAVOR_GNU(PROGRAM-PATH, IF-SUCCESS, [IF-FAILURE]) # copied autoconf internal _AC_PATH_PROG_FLAVOR_GNU -m4_define([BIRD_CHECK_PROG_FLAVOR_GNU], +AC_DEFUN([BIRD_CHECK_PROG_FLAVOR_GNU], [# Check for GNU $1 case `"$1" --version 2>&1` in *GNU*) @@ -143,4 +143,4 @@ m4_ifval([$3], [*) $3;; ])esac -])# +]) -- 2.12.0

1 0

[PATCH 01/13] aclocal.m4: remove unused macro
by Ruben Kerkhof 18 Mar '17

18 Mar '17

--- aclocal.m4 | 36 ------------------------------------ 1 file changed, 36 deletions(-) diff --git a/aclocal.m4 b/aclocal.m4 index f2775422..a13b71f3 100644 --- a/aclocal.m4 +++ b/aclocal.m4 @@ -34,42 +34,6 @@ for size in 1 2 4 8; do done ]) -dnl BIRD_CHECK_ENDIAN is unused and obsolete -AC_DEFUN([BIRD_CHECK_ENDIAN], -[AC_CACHE_CHECK([CPU endianity], bird_cv_c_endian,[ -AC_TRY_RUN([ -#include <stdio.h> - -unsigned int x = 0x12345678; -unsigned char *z = (unsigned char *) &x; - -int main(void) -{ - FILE *f = fopen("conftestresult", "w"); - if (!f) return 10; - fprintf(f, "%02x %02x %02x %02x", *z, *(z+1), *(z+2), *(z+3)); - fclose(f); - exit(0); -} -],[ - endian=`cat conftestresult` - if test "$endian" = "12 34 56 78" ; then - bird_cv_c_endian=big-endian - elif test "$endian" = "78 56 34 12" ; then - bird_cv_c_endian=little-endian - fi -],[endian="test program failed"],[endian="not available, we're cross compiling"]) -if test -z "$bird_cv_c_endian" ; then - AC_MSG_RESULT($endian) - AC_MSG_ERROR([Cannot determine CPU endianity.]) - fi -]) -case $bird_cv_c_endian in - big-endian) AC_DEFINE(CPU_BIG_ENDIAN) ;; - little-endian) AC_DEFINE(CPU_LITTLE_ENDIAN) ;; - esac -]) - AC_DEFUN([BIRD_CHECK_STRUCT_ALIGN], [AC_CACHE_CHECK([usual alignment of structures],bird_cv_c_struct_align,[ AC_TRY_RUN([ -- 2.12.0

1 0

cppcheck on bird-1.6.3
by Илья Шипицин 16 Mar '17

16 Mar '17

Hello, I wonder about those issues: [sysdep/unix/krt.c:1189]: (error) Memory is allocated but not initialized: n [sysdep/unix/krt.c:1190]: (error) Memory is allocated but not initialized: n Cheers, Ilya Shipitsin

1 0

Multi core validate ospf
by Fernando Galvão 15 Mar '17

15 Mar '17

Hello guys, I wanted your help to validate if the bird with ospf is running with multi thread. I have a dell r410 server with 2 x56 quad core xeon processors totaling 16 cores. This server was serving pppoe server with accel-ppp + quagga ospf v4 / v6 with route summarization. With 2,000 sessions and going on around 1GB, the quagga process blew the cpu and all sessions were disconnected. The problem is that the quagga is single core so it uses only one core. Looking, I've seen that the bird is multi-core. I started to run, I have 2,200 pppoe sessions but still early to get high traffic, I'm passing 300MB and the bird has peaks in a 25% core. I would like to know if I have validate that it is distributing the process between CPUs? Or if you have any configuration for this. Fernando Galvão Wantel Telecom +55 87 3866-5200

4 4

Some autoconf cleanups
by Ruben Kerkhof 14 Mar '17

14 Mar '17

The goal of this series is to make it possible to run autoreconf on a git checkout. It does this by moving configure.in to configure.ac and fixes deprecation warnings and errors generated by autoheader. I have some follow up patches with more cleanups, but this series is already large enough as it is.

2 16

[PATCH,RFC 0/3] Secondary remote AS support for protocol BGP.
by Lennert Buytenhek 14 Mar '17

14 Mar '17

Hi! These patches implement a way of simplifying BGP topology migrations that involve changing AS numbers on one or both ends of BGP sessions. The first patch allows a BGP peer to connect to us using either of two different AS numbers specified in our local config. (See the commit message for that patch for more details.) The second patch fixes what I think is a reconfiguration bug for protocol BGP and is in the stack mainly because of patch three. The third patch avoids bouncing a BGP session when reconfiguring its BGP protocol if all we're changing is adding a secondary remote AS to the protocol or changing one of two remote AS numbers that is not currently being advertised by the peer. (Again, see the commit message for that patch for more details.) Any feedback appreciated! And even if you happen not to like patch 3, I hope that you'd still be willing to consider patch 1 (and 2). Cheers, Lennert

3 13