I’ll look into the bgp_agent_map.
In this case, both route servers *should* contain the same routing information
however they will never be perfectly symmetrical.
I’ll peer with the “primary” RS for now but I’ll be looking into a GoBGP daemon
to feed BMP data into pmacct instead of direct peer.
Also, is there a way to feed BMP and flow data into pmacct off a Kafka bus? It
would be good to buffer input data streams into pmacct to allow for better data
buffering and system resiliency?
> On 2 Feb 2018, at 11:48 pm, Paolo Lucente <pa...@pmacct.net> wrote:
> Hi Tim,
> Thanks for your email. Yes, you can map a BGP speaker to a flow exporter
> via bgp_agent_map; since you speak of route servers, that is plural,
> would they all return an omogeneous view in term of routes and more of
> them are in place for redundancy? Or routes would be different at each
> of them and you would be looking for some best-path computation being
> done in pmacct (which is not the case today)?
> On Thu, Feb 01, 2018 at 02:45:16PM +1100, Tim Raphael wrote:
>> Hi All!
>> I work for an IXP and we’re attempting to collect flows across our fabric
>> however I’m not sure how to address our architecture.
>> We run L2 PE devices that don’t interact with L3 at all as far as peers and
>> peering traffic are concerned - they’re just L2 boxes.
>> Peers connect with BGP to a pair of route servers for BGP.
>> Now, I understand the pmacct can do prefix lookup in BGP but I’m under the
>> assumption that the sFlow peer_src_ip has to match the BGP peer source
>> address to a flow to be decorated with BGP specific information. Given our
>> flows will be coming from our PEs and BGP from our route servers - is this
>> possible in pmacct?
>> - Tim
>> pmacct-discussion mailing list
> pmacct-discussion mailing list
pmacct-discussion mailing list