Hi, We also use an external script to generate our BIRD configs. I can recommend using Python with Jinja2 templates. Something like this: Python calling jinja template: templateVars = { "peer_asn" : 1234 "peer_id" : 12 "ips" : ["1.1.1.1", "1.1.1.2"], } templateEnv = jinja2.Environment(loader=jinja2.FileSystemLoader(searchpath=os.path.join("/"))) template = templateEnv.get_template(template.tmp) rendr = template.render( templateVars ) filename = config["outputs"]["rt_tables"] + "rt_tables" f = open(filename, 'w') f.write(rendr.encode('utf8')) f.close() Jinja template: define ips_AS{{ peer.asn }}_PeerID{{ peer.id }} = [ {% for ip in ips %} {% if loop.last %} {{ ip }} {% else %} {{ ip }}, {% endif %} {% endfor %} ]; and the like Liran said just do a birdc configure On 14.02.2017 13:20, Stanislaw wrote:

Hi,

Yes, you can do that. The best practice (at least for myself) is using the separate auto-generated config with prefix-lists which is included to the bird.conf. After regenerating it via some of your scripts, just run birdc configure. The reconfiguration of import/export filters is completely seamless. As for the performance impact: it uses some extra CPU power for recalculating bestpathes but it lasts for ~1 second on modern CPUs.

Whant kind of example do you want? Just make another file, write your prefix-lists as a variables (or constants) there and use the 'include' statement in the bird.conf. Ofc, use those variables in your filters.

Liran Schour писал 2017-02-14 13:46:

...

Hi all,

I want to implement a dynamic filter where I will export a route based on matching against a dynamic set of prefixes. The set is maintained according to a database.

1. Can I maintain a dynamic set of prefixes and change the set on the fly? What are the performance impacts of this? 2.. Where can I find such an example?

- Liran