bgpq4 - Man Page

The bgpq4 utility used to generate configurations (prefix-lists, extended access-lists, policy-statement terms and as-path lists) based on IRR data.

The options are as follows:

-4

generate IPv4 prefix/access-lists (default).

-6

generate IPv6 prefix/access-lists (IPv4 by default).

-A

try to aggregate prefix-lists as much as possible (not all output formats supported).

-a asn

specify what asn shall be denied in case of empty prefix-list (OpenBGPD)

-B

generate output in OpenBGPD format (default: Cisco)

-b

generate output in BIRD format (default: Cisco).

-d

enable some debugging output.

-e

generate output in Arista EOS format (default: Cisco).

-E

generate extended access-list (Cisco), policy-statement term using route-filters (Juniper), [ip|ipv6]-prefix-list (Nokia) or prefix-sets (OpenBGPd).

-f number

generate input as-path access-list.

-F fmt

generate output in user-defined format.

-G number

generate output as-path access-list.

-H number

generate output as-list for as-path-origin filter (JunOS 21.3R1+)

-h host[:port]

host running IRRD database (default: rr.ntt.net).

-J

generate config for Juniper (default: Cisco).

-j

generate output in JSON format (default: Cisco).

-K

generate config for Mikrotik ROSv6 (default: Cisco).

-K7

generate config for Mikrotik ROSv7 (default: Cisco).

-l name

name of generated entry.

-L limit

limit recursion depth when expanding as-sets.

-m len

maximum prefix-length of accepted prefixes (default: 32 for IPv4 and 128 for IPv6).

-M match

extra match conditions for Juniper route-filters.

-n

generate config for Nokia SR OS MD-CLI (Cisco IOS by default)

-n2

generate config for Nokia SR Linux (Cisco IOS by default)

-N

generate config for Nokia SR OS classic CLI (Cisco IOS by default).

-p

emit prefixes where the origin ASN is 23456 or in the private ASN range (disabled by default).

-r len

allow more specific routes starting with specified masklen too.

-R len

allow more specific routes up to specified masklen too.

-s

generate sequence numbers in IOS-style prefix-lists.

-S sources

use specified sources only (recommended: RPKI,AFRINIC,APNIC,ARIN,LACNIC,RIPE).

-t

generate as-sets for OpenBGPd, BIRD and JSON formats.

-T

disable pipelining (not recommended).

-U

generate config for Huawei devices (Cisco IOS by default)

-u

generate config for Huawei devices in XPL format (Cisco IOS by default)

-W len

generate as-path strings of no more than len items (use 0 for infinity).

-X

generate config for Cisco IOS XR devices (plain IOS by default).

-z

generate route-filter-lists (JunOS 16.2+).

OBJECTS

means networks (in prefix format), autonomous systems, as-sets and route-sets.

EXCEPT OBJECTS

those objects will be excluded from expansion.

Generating named juniper prefix-filter for AS20597:

$ bgpq4 -Jl eltel AS20597 policy-options { replace: prefix-list eltel { 81.9.0.0/20; 81.9.32.0/20; 81.9.96.0/20; 81.222.128.0/20; 81.222.192.0/18; 85.249.8.0/21; 85.249.224.0/19; 89.112.0.0/19; 89.112.4.0/22; 89.112.32.0/19; 89.112.64.0/19; 217.170.64.0/20; 217.170.80.0/20; } }

For Cisco we can use aggregation (-A) flag to make this prefix-filter more compact:

$ bgpq4 -Al eltel AS20597 no ip prefix-list eltel ip prefix-list eltel permit 81.9.0.0/20 ip prefix-list eltel permit 81.9.32.0/20 ip prefix-list eltel permit 81.9.96.0/20 ip prefix-list eltel permit 81.222.128.0/20 ip prefix-list eltel permit 81.222.192.0/18 ip prefix-list eltel permit 85.249.8.0/21 ip prefix-list eltel permit 85.249.224.0/19 ip prefix-list eltel permit 89.112.0.0/18 ge 19 le 19 ip prefix-list eltel permit 89.112.4.0/22 ip prefix-list eltel permit 89.112.64.0/19 ip prefix-list eltel permit 217.170.64.0/19 ge 20 le 20

Prefixes 89.112.0.0/19 and 89.112.32.0/19 now aggregated into single entry 89.112.0.0/18 ge 19 le 19.

Well, for Juniper we can generate even more interesting policy-options, using -M <extra match conditions>, -R <len> and hierarchical names:

$ bgpq4 -AJEl eltel/specifics -r 29 -R 32 -M "community blackhole" AS20597 policy-options { policy-statement eltel { term specifics { replace: from { community blackhole; route-filter 81.9.0.0/20 prefix-length-range /29-/32; route-filter 81.9.32.0/20 prefix-length-range /29-/32; route-filter 81.9.96.0/20 prefix-length-range /29-/32; route-filter 81.222.128.0/20 prefix-length-range /29-/32; route-filter 81.222.192.0/18 prefix-length-range /29-/32; route-filter 85.249.8.0/21 prefix-length-range /29-/32; route-filter 85.249.224.0/19 prefix-length-range /29-/32; route-filter 89.112.0.0/17 prefix-length-range /29-/32; route-filter 217.170.64.0/19 prefix-length-range /29-/32; } } } }

generated policy-option term now allows all specifics with prefix-length between /29 and /32 for eltel networks if they match with special community blackhole (defined elsewhere in configuration).

Of course, this version supports IPv6 (-6):

$ bgpq4 -6l as-retn-6 AS-RETN6 no ipv6 prefix-list as-retn-6 ipv6 prefix-list as-retn-6 permit 2001:7fb:fe00::/48 ipv6 prefix-list as-retn-6 permit 2001:7fb:fe01::/48 [....]

and assumes your device supports 32-bit ASNs

$ bgpq4 -Jf 112 AS-SPACENET policy-options { replace: as-path-group NN { as-path a0 "^112(112)*$"; as-path a1 "^112(.)*(1898|5539|8495|8763|8878|12136|12931|15909)$"; as-path a2 "^112(.)*(21358|23456|23600|24151|25152|31529|34127|34906)$"; as-path a3 "^112(.)*(35052|41720|43628|44450|196611)$"; } }

see `AS196611` in the end of the list ? That's a 32-bit ASN.

If you want to generate configuration not for routers, but for some other programs/systems, you may use user-defined formatting, like in example below:

$ bgpq4 -F "ipfw add pass all from %n/%l to any\n" as3254 ipfw add pass all from 62.244.0.0/18 to any ipfw add pass all from 91.219.29.0/24 to any ipfw add pass all from 91.219.30.0/24 to any ipfw add pass all from 193.193.192.0/19 to any

Recognized format sequences are:

Please note that no new lines are inserted automatically after each sentence. You have to add them into format string manually, otherwise the output will be in one single line (sometimes it makes sense):

$ bgpq4 -6F "%n/%l; " as-eltel 2001:1b00::/32; 2620:4f:8000::/48; 2a04:bac0::/29; 2a05:3a80::/48;

By default bgpq4 trusts data from all the databases mirrored into NTT's IRR service. Unfortunately, not all these databases are equal in how much we can trust their data. RIR maintained databases (AFRINIC, ARIN, APNIC, LACNIC and RIPE) shall be trusted more than the others because they have the knowledge about who the rightful holders of resources are, while other databases lack this knowledge and can (and, actually do) contain stale data: no one but the RIRs care to remove outdated route-objects when address space is de-allocated or transferred. In order to keep their filters both compact and actual, bgpq4 users are encouraged to use '-S' flag to limit database sources to only the ones they trust.

General recommendations:

Use a minimal set of RIR databases (only those in which you and your customers have registered route-objects).

Avoid using ARIN-NONAUTH and RIPE-NONAUTH as trusted sources: these records were created in the database, but for address space allocated to different RIRs, so the NONAUTH databases have no chance to confirm validity of the route objects they contain.

$ bgpq4 -S RIPE,RADB as-space
no ip prefix-list NN
ip prefix-list NN permit 195.190.32.0/19

$ bgpq4 -S RADB,RIPE as-space
no ip prefix-list NN
ip prefix-list NN permit 45.4.4.0/22
ip prefix-list NN permit 45.4.132.0/22
ip prefix-list NN permit 45.6.128.0/22
ip prefix-list NN permit 45.65.184.0/22
[...]

To improve `bgpq4` performance when expanding extra-large AS-SETs you shall tune OS settings to enlarge TCP send buffer.

FreeBSD can be tuned in the following way:

sysctl -w net.inet.tcp.sendbuf_max=2097152

Linux can be tuned in the following way:

sysctl -w net.ipv4.tcp_window_scaling=1

sysctl -w net.core.rmem_max=2097152

sysctl -w net.core.wmem_max=2097152

sysctl -w net.ipv4.tcp_rmem="4096 87380 2097152"

sysctl -w net.ipv4.tcp_wmem="4096 65536 2097152"

This project uses autotools. If you are building from the repository, run the following command to prepare the build system:

./bootstrap

In order to compile the software, run:

./configure

make

make install

If you wish to remove the generated build system files from your working tree, run:

make maintainer-clean

In order to create a distribution archive, run:

make dist

When everything is OK, bgpq4 generates access-list to standard output and exits with status == 0. In case of errors they are printed to stderr and program exits with non-zero status.

Alexandre Snarskii, Christian David, Claudio Jeker, Job Snijders, Massimiliano Stucchi, Michail Litvak, Peter Schoenmaker, Roelf Wichertjes, and contributions from many others.

https://github.com/bgp/bgpq4 BGPQ4 on Github.

http://bgpfilterguide.nlnog.net/ NLNOG's BGP Filter Guide.

https://tcp0.com/cgi-bin/mailman/listinfo/bgpq4 Users and interested parties can subscribe to the BGPQ4 mailing list bgpq4@tcp0.com

Job Snijders ⟨job@sobornost.net⟩