- Bird-users - Mailing Lists

Allowing multiple instances of kernel protocol to use same table
by Alexander Demenshin 18 Aug '14

18 Aug '14

Hi, Is there any specific reason why each instance of kernel protocol must use unique table? What I want to achieve is to export different protocols (OSPF, BGP etc) to different kernel tables, but the only way I can do this currently is to use pipe, instead of export filter only. My concern is duplication of data - BGP table has over 500k routes, so keeping this in master *and* piping to another one for kernel export is a waste of resources (both RAM and CPU). Thank you! -- With best regards, Alexander.

3 5

Removing device-bound routes when interface is going down
by Alexander Demenshin 18 Aug '14

18 Aug '14

Hi (again), I've encountered a problem which is (most likely) is caused by a bug in the Linux kernel, but significantly affects routing handling in bird (everything else like quagga is most likely also affected). When route is added manually/externally to the kernel, and bound to specific interface, like this: $ ip route add 10.1.1.0/24 dev eth0 ...it is correctly identified by bird (via async netlink message), but when the interface goes down, this route is not deleted, as there is no corresponding netlink message (tested on most recent kernel, 3.16.1, and also on older 3.11, this happens only on if-down event, manual deletion works fine). Since bird does not see the route removal, it is still kept in the routing table, thus, it is still announced via OSPF (in my particular case), however the kernel knows nothing about this. Obviously, it is possible to set up periodic scanning of kernel table with small interval, but even then, this removal will not be instant, and additionally, it introduces additional CPU usage - even if the table in question is small by itself, but there is another huge table, this scan will consume significant time, in my case - it takes 0.3s to dump main table (~10 entries), all due to huge BGP table (~500k entries), thus scanning every 10 seconds will consume 3% CPU (mostly for nothing, as this kind of changes are relatively seldom). Fixing this problem in the mainstream kernel (if this is indeed unintentional behavior) will clearly take ages, but it could be easily solved in bird itself - if interface is going down, all routes pointing to this device must be removed (not only direct and static routes - where it works perfectly). I feel that this workaround is quite easy to implement, but not sure where exactly to look, thus I'll be really grateful for right directions :) Thank you! -- With best regards, Alexander.

4 6

BGP Filtering question
by Eloi Carbó Solé 18 Aug '14

18 Aug '14

Hi everyone, I am trying to do something that I think that should be so easy but I am not success yet. In IPv4 I need to know if there is any possibility to execute OS commands and receive its output (in filters or functions) owing to receive the list of dev names to accept (concretely in a OpenWRT shell or similar). An example of the routes I want to accept is: 10.201.192.0/24 dev bmxOut_GSoC-0a proto static metric 1024 Otherwise, my other possiblity is to use statically the device name: bmx_____, using the pattern "bmx*" to filter them (it is possible?), but I am not really sure if there is a "dev" option, if I may use "ifname" (it is outgoing iface, so I suppose that no) or something like to check it. Best regards, -- Eloi Carbó

2 1

L2 MTU issues - is BFD proper solution?
by Bartosz Radwan 15 Aug '14

15 Aug '14

We experienced recently several times issues caused by eg. EAPS path swich in one of our L2 transmission providers (new path MTU was not big enough due to jumbo frames not enabled - that's what they told us, but in fact it does not meter - it's out of our control). From our point of view this situation sucks - we have several BGP sessions, terminated on 2 different BGP routers in different data centers, but problems as descripbed above in just one L2 provider network may cause our all services are not fully operational - especially when afeccted BGP sessions provides best routes for some popular services like FB, YT or… redtube.com ;). So, the question: is BFD able to detect broken L2 MTU and if detected - disable affected BGP sessions? If not BFD, any other ideas? It's possible to detect L2 problems outside bird and then disable protocols, but this may be complicated and error prone, so in fact - i do not want to to that. -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? Bartosz Radwan

4 3

Re: Multihop BGP cannot use direct gateway mode - misidentification of multihop sessions?
by Alex Bligh 13 Aug '14

13 Aug '14

Any ideas on the below? Alex On 8 Aug 2014, at 20:05, Alex Bligh <alex(a)alex.org.uk> wrote: > When I start bird using 1.4.0-1 (Ubuntu Trusty) I see the error: > > warnings.log:Aug 8 13:47:03 node-10-157-128-41 bird: /etc/bird/bird.conf, line 62: Multihop BGP cannot use direct gateway mode > > This error did not occur on 1.3.4-1 with the same config on Ubuntu Precise. > > The config I am using is pasted below, as is the output of 'ip addr show'. > > Line 62 of the config is the closing bracket of a BGP protocol definition. The neighbour statement is on line 29: > neighbor 10.157.192.1 as 65000 > > You will note that interface evrr-000000 has IP address 10.157.192.41/20, and the peer is IP address 10.157.192.1. This should thus not be a multihop session. The output of 'ip route show' (also below) confirms this. > > It is an iBGP session (in that both ends have the same address) but I am deliberately hoping to use gateway direct, i.e. want to put the advertised next hop straight into the RIB without doing a recursive lookup. Specifically if something else on the same /20 gets advertised as a next hop, I still want to send the traffic to the advertising router. > > The documentation says: > >> For received routes, their gw (immediate next hop) attribute is computed from received bgp_next_hopattribute. This option specifies how it is computed. Direct mode means that the IP address frombgp_next_hop is used if it is directly reachable, otherwise the neighbor IP address is used. Recursive mode means that the gateway is computed by an IGP routing table lookup for the IP address from bgp_next_hop. Recursive mode is the behavior specified by the BGP standard. Direct mode is simpler, does not require any routes in a routing table, and was used in older versions of BIRD, but does not handle well nontrivial iBGP setups and multihop. Recursive mode is incompatible with sorted tables. Default: direct for direct sessions, recursive for multihop sessions. > > > From this, gateway direct should be the default for a direct session in any case. It would appear to me that bird is somehow misidentifying the session as a multihop session. > > Removing 'gateway direct' fixes the error (but obviously not the behaviour). > > This is running in an network container if that's of any interest. > > Any ideas? > > (apologies for a potential dupe - I sent this to what I think must have been an old list email address) > > -- > Alex Bligh > > > > root@node-10-157-128-41:/# ip addr show > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 > inet 127.0.0.1/8 scope host lo > valid_lft forever preferred_lft forever > inet6 ::1/128 scope host > valid_lft forever preferred_lft forever > 2: dummy2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000 > link/ether 5a:17:eb:e4:7b:10 brd ff:ff:ff:ff:ff:ff > inet6 fe80::5817:ebff:fee4:7b10/64 scope link > valid_lft forever preferred_lft forever > 3: dummy: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000 > link/ether 12:7f:06:13:e8:91 brd ff:ff:ff:ff:ff:ff > inet 10.99.99.1/24 scope global dummy > valid_lft forever preferred_lft forever > inet6 fe80::107f:6ff:fe13:e891/64 scope link > valid_lft forever preferred_lft forever > 12: evrr-000000: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000 > link/ether ee:19:d3:6f:24:1b brd ff:ff:ff:ff:ff:ff > inet 10.157.192.41/20 scope global evrr-000000 > valid_lft forever preferred_lft forever > inet6 fc00:41c8:10:1dd::29/64 scope global > valid_lft forever preferred_lft forever > inet6 fe80::ec19:d3ff:fe6f:241b/64 scope link > valid_lft forever preferred_lft forever > > root@node-10-157-128-41:/# ip route show > 10.99.99.0/24 dev dummy proto kernel scope link src 10.99.99.1 > 10.157.192.0/20 dev evrr-000000 proto kernel scope link src 10.157.192.41 > > > 1 # Configure logging > 2 log syslog { debug, trace, info, remote, warning, error, auth, fatal, bug }; > 3 log stderr all; > 4 > 5 router id 10.157.192.41; > 6 > 7 # Import kernel routes. These come in with RTS_INHERIT > 8 protocol kernel evrkernel { > 9 persist; # Don't remove routes on bird shutdown > 10 scan time 10; # Scan kernel routing table every 20 seconds > 11 import all; > 12 export all; > 13 learn; # Learn all alien routes from the kernel > 14 } > 15 > 16 # This pseudo-protocol watches all interface up/down events. > 17 protocol device { > 18 scan time 10; # Scan interfaces every 10 seconds > 19 } > 20 > 21 # Import all directly connected routes. These come in with RTS_DEVICE > 22 protocol direct evrdirect { > 23 interface "-evrr-000000", "evrr-*"; > 24 export all; > 25 } > 26 > 27 protocol bgp evrbgp1 { > 28 local as 65000; > 29 neighbor 10.157.192.1 as 65000; > 30 > 31 gateway direct; > 32 export filter { > 33 # RTS_INHERIT means routers from the kernel > 34 if net = 10.157.192.0/20 then reject; > 35 if net = 0.0.0.0/8 then reject; > 36 if (source = RTS_INHERIT) || (source = RTS_DEVICE) then { > 37 # Add no-export > 38 bgp_community.add((65535, 65281)); > 39 # Add no-advertise > 40 bgp_community.add((65535, 65282)); > 41 > 42 bgp_local_pref = 100; > 43 > 44 accept; > 45 } > 46 reject; > 47 }; > 48 import filter { > 49 if net = 0.0.0.0/0 then accept; > 50 reject; > 51 }; > 52 next hop self; > 53 > 54 # timers > 55 hold time 12; > 56 startup hold time 12; > 57 keepalive time 4; > 58 connect retry time 6; > 59 start delay time 1; > 60 error wait time 2, 16; > 61 error forget time 16; > 62 } > 63 > 64 protocol bgp evrbgp2 { > 65 local as 65000; > 66 neighbor 10.157.192.2 as 65000; > 67 > 68 gateway direct; > 69 export filter { > 70 # RTS_INHERIT means routers from the kernel > 71 if net = 10.157.192.0/20 then reject; > 72 if net = 0.0.0.0/8 then reject; > 73 if (source = RTS_INHERIT) || (source = RTS_DEVICE) then { > 74 # Add no-export > 75 bgp_community.add((65535, 65281)); > 76 # Add no-advertise > 77 bgp_community.add((65535, 65282)); > 78 > 79 bgp_local_pref = 100; > 80 > 81 accept; > 82 } > 83 reject; > 84 }; > 85 import filter { > 86 if net = 0.0.0.0/0 then accept; > 87 reject; > 88 }; > 89 next hop self; > 90 > 91 # timers > 92 hold time 12; > 93 startup hold time 12; > 94 keepalive time 4; > 95 connect retry time 6; > 96 start delay time 1; > 97 error wait time 2, 16; > 98 error forget time 16; > 99 } > 100 > -- Alex Bligh

3 6

dmalloc
by Nic Cave-Lynch 06 Aug '14

06 Aug '14

Hi people I'm working on a custom routing module and I'm having problems with memory allocation. I've tried compiling with --enable-memcheck, having installed libdmalloc (with apt-get install libdmalloc-dev libdmalloc5). make now fails as below. In lib/resource.h there's a check for libdmalloc which then refers to _xFOO_leap functions which I guess were once included with libdmalloc, but they're not there now. Is there a way of using libdmalloc, (or some other equivalent) that is currently operational? Thanks Nic C-L make -C obj all make[1]: Entering directory `/home/nic/projects/bird/obj' set -e ; for a in lib conf ; do make -C $a subdir ; done make[2]: Entering directory `/home/nic/projects/bird/obj/lib' gcc -I../ -I../.. -g -O2 -pthread -Wall -Wstrict-prototypes -Wno-parentheses -Wno-pointer-sign -fno-strict-aliasing -fno-strict-overflow -o io.o -c io.c io.c: In function ‘sk_alloc_bufs’: io.c:988:5: warning: implicit declaration of function ‘_xmalloc_leap’ [-Wimplicit-function-declaration] s->rbuf = s->rbuf_alloc = xmalloc(s->rbsize); ^ io.c:988:29: warning: assignment makes pointer from integer without a cast [enabled by default] s->rbuf = s->rbuf_alloc = xmalloc(s->rbsize); ^ io.c:991:29: warning: assignment makes pointer from integer without a cast [enabled by default] s->tbuf = s->tbuf_alloc = xmalloc(s->tbsize); ^ io.c: In function ‘sk_free_bufs’: io.c:1000:5: warning: implicit declaration of function ‘_xfree_leap’ [-Wimplicit-function-declaration] xfree(s->rbuf_alloc); ^ io.c: In function ‘sk_set_rbsize’: io.c:1043:17: warning: assignment makes pointer from integer without a cast [enabled by default] s->rbuf_alloc = xmalloc(val); ^ io.c: In function ‘sk_set_tbsize’: io.c:1058:56: error: macro "xrealloc" passed 2 arguments, but takes just 1 s->tbuf = s->tbuf_alloc = xrealloc(s->tbuf_alloc, val); ^ io.c:1058:29: error: ‘xrealloc’ undeclared (first use in this function) s->tbuf = s->tbuf_alloc = xrealloc(s->tbuf_alloc, val); ^ io.c:1058:29: note: each undeclared identifier is reported only once for each function it appears in io.c: In function ‘io_loop’: io.c:1970:12: warning: variable ‘e’ set but not used [-Wunused-but-set-variable] int e; -- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

2 1

next-hop announcement/withdrawal process
by Pavlos Parissis 04 Aug '14

04 Aug '14

Hi, I am building a solution where I have N HAProxy servers which have BGP and BFD sessions with 2 upstream routers. Bird runs on all HAProxy servers and announces the same prefixes via BGP. I am using ECMP on upstream routes to balance incoming traffic to these HAProxy servers, which balance traffic to N backend servers. I need to advertise/withdraw several IP prefixes within the same range (192.168.200.0/24) based on the health of a service, in this case the frontend on HAProxy. These IPs are assigned only the loopback interface. The problem I am facing is that when I adjust a filter in order to withdraw the route, all routes are withdrawn and the remaining are announced again. According to the doc if a reload on bird brings a new configuration for a protocol, that protocol will be restarted. So, it is expected behavior. Below is the log[1] after I removed '192.168.200.1' from prefix set anycast_prefixes in loopback_ACL filter and reloaded bird. I was monitoring BGP protocols in a loop running bird show protocols all "bgp8" and saw that 'export updates' counter was increased from 2 to 3 and 'export withdraws' counter was increased from 0 to 2. Is there a way to withdraw to a route from a list without effecting the rest routes from the list? The only way I have found so far, it was by removing the specific IP from the loopback interface. But, this solution prevents a reload/restart on HAProxy as it can't bind anymore to the socket Cheers, Pavlos [1] 23:37:52 <INFO> Reconfiguration requested by SIGHUP 23:37:52 <INFO> Reconfiguring 23:37:52 <TRACE> direct1: Reconfigured 23:37:52 <INFO> Reloading protocol direct1 23:37:52 <INFO> Restarting protocol direct1 23:37:52 <TRACE> direct1: Shutting down 23:37:52 <TRACE> direct1: State changed to flush 23:37:52 <TRACE> bgp_peer1: Reconfigured 23:37:52 <TRACE> bgp_peer2: Reconfigured 23:37:52 <INFO> Reconfigured 23:37:52 <TRACE> direct1 > removed [sole] 192.168.200.1/32 dev lo 23:37:52 <TRACE> direct1 > removed [sole] 192.168.200.2/32 dev lo 23:37:52 <TRACE> direct1: State changed to down 23:37:52 <TRACE> direct1: Starting 23:37:52 <TRACE> direct1: Connected to table master 23:37:52 <TRACE> direct1: State changed to feed 23:37:52 <TRACE> direct1 < primary address 192.168.200.1/32 on interface lo added 23:37:52 <TRACE> direct1 > filtered out 192.168.200.1/32 dev lo 23:37:52 <TRACE> direct1 < secondary address 127.0.0.0/8 on interface lo added 23:37:52 <TRACE> direct1 < secondary address 192.168.200.2/32 on interface lo added 23:37:52 <TRACE> direct1 > added [best] 192.168.200.2/32 dev lo 23:37:52 <TRACE> bgp_peer2 < filtered out 192.168.200.2/32 dev lo 23:37:52 <TRACE> bgp_peer1 < filtered out 192.168.200.2/32 dev lo 23:37:52 <TRACE> direct1 < primary address 192.168.80.0/20 on interface eth0 added 23:37:52 <TRACE> direct1: State changed to up 23:37:54 <TRACE> bgp_peer2: Sending KEEPALIVE 23:37:54 <TRACE> bgp_peer1: Sending KEEPALIVE 23:37:54 <TRACE> bgp_peer2: Got KEEPALIVE 23:37:56 <TRACE> bgp_peer1: Got KEEPALIVE bird.conf # Override router ID router id 192.168.88.194; # Listens only on eth0's IP listen bgp address 192.168.88.194; filter loopback_ACL prefix set anycast_prefixes; { anycast_prefixes = [ 192.168.200.1/32, 192.168.200.2/32 ]; if net ~ anycast_prefixes then accept; else reject; } filter anycast_range { if net ~ [ 192.168.200.0/24 ] then accept; else reject; } protocol direct { interface "lo"; debug all; import filter loopback_ACL; export none; } protocol kernel kernel1 { disabled yes; } protocol device { scan time 10; # TODO do I need this? } protocol static { disabled yes; } protocol bfd { interface "eth0" { min rx interval 500 ms; min tx interval 500 ms; idle tx interval 1000 ms; multiplier 3; }; } template bgp BGP_PEERS { bfd on; debug all; import none; export filter anycast_range; direct; hold time 30; startup hold time 240; connect retry time 120; keepalive time 10; start delay time 5; error wait time 60, 300; error forget time 300; disable after error off; next hop self; path metric 1; default bgp_med 0; default bgp_local_pref 0; } protocol bgp bgp_peer1 from BGP_PEERS { disabled no; description "My-BGP-Peer1"; local as 46111; neighbor 192.168.95.252 as 46111; source address 192.168.88.194; } protocol bgp bgp_peer2 from BGP_PEERS { disabled no; description "My-BGP-Peer2"; local as 46111; neighbor 192.168.95.253 as 46111; source address 192.168.88.194; }

5 8

Allow configuration of BGP sessions to non standard ports.
by João Taveira Araújo 02 Aug '14

02 Aug '14

Hi, The current configuration options assume that one or both BGP peers will be listening on port 179. As a result it becomes impossible to configure BIRD instances to talk to each other if both are running BGP on non standard ports, or if a single instance is on a non standard port and running in passive mode. I've attached a patch which I think does the trick, allowing the user to specify a 'neighbor port <portnum>;' option, but I'm not that familiar with the codebase so I might be missing something. If the patch is not appropriate in its current form, please consider this a feature request for future versions. Thanks, Joao.

1 0

Prefix count limit?
by Tom van Leeuwen 31 Jul '14

31 Jul '14

Hi everyone, I got an announcement from my iptransit provider stating the ipv4 routing table size has now exceeded 500k routes, so I thought lets check if bird is able to handle this and actually find out if there are any limitations. I could not find it on the website, but if I remember correctly the amount of routes is only impacted by the amount of memory the bird server has? Kind regards, Tom van Leeuwen

3 2

how to use "Operators" in bird
by 曾小小 30 Jul '14

30 Jul '14

hi, My system is Ubuntu 13.04, kernel version is 3.8.0-21-generic, bird version is 1.4.4. Device configuration is as follows: 1. about function function rt_import (int asn; int set peer_asns; prefix set peer_nets) { if (net ~ peer_nets) then return false!; } 2. about filter filter bgp_upstream { if rt_import (64609, [64609], [192.168.1.0/24]) then reject; } In the above function, I use net ~ peer_nets to filter out peer_nets route, but in filter medium, peer_nets defined as 192.168.1.0/24, which would lead to such a fine route 192.168.1.1 will be filtered out. If the test using the "-" operator, will lead filter to filter out all the routes. Why? In the above example, how can I use "operators" to achieve peer_nets defined as 192.168.1.0/24, 192.168.1.1 subnet address to ensure it will not be filtered out. Thank you very much for nothing always help, thanks.

2 1