MPLS L3VPN customer aggregate-address with route-map

[fsebera]

I need to create aggregate-addresses under MPLS L3VPN CE filter with a route-map filter  - 

various options appear to be available including

--snip--

• as-confed-set
• as-set
• attribute-map
• route-map

--snip--

ex: I type in the following:

address-family ipv4 vrf MPLS_VPN_C800

aggregate-address <net> <mask> route-map CUSTOMER-800

.... and this is what I get:

do sh run

address-family ipv4 vrf MPLS_VPN_C800

aggregate-address <net> <mask> attribute-map CUSTOMER-800

I have to admit I haven't used aggregate-address under L3 VPN before but it seems I must be missing something or just not understanding. I performed a google search and found nothing useful. How do I get to use the route-map option?

Thank you

Frank

[omz]

Hi 

What is the route-map used for? To filter routes? I think .. supress/unsupress maps are used to filter routes with aggregate-address command.

I think .. its changing to attribute map because its expecting the route-map to manipulate the BGP attributes and filter routes. 

[fsebera]

Hi guys,

When I type in

router bgp #

address-family ipv4 vrf L3VPN-CUSTOMER-800

aggregate-address <address> <mask> route-map <RM_name>

with the intention of filtering prefixes and the operating system -IOS-XE- automatically changes the words "route-map" to "attribute-map" even though route-map is a valid option, are you suggesting this is cosmetic and I should not worry about it?

Also, I am only trying to build the aggregate (summary) address with BGP prefixes. If a route exist in the table but it is not a BGP derived prefix the filter must ignore it - reasons are not mine but customers.

Thanks again

Frank

[Giuseppe Larosa]

Hello Frank,

 the correct option with aggregate-address to make it advertised conditioned should be advertise-map <route-map-name>.

>> Also, I am only trying to build the aggregate (summary) address with BGP prefixes. If a route exist in the table but it is not a BGP derived prefix the filter must ignore it - reasons are not mine but customers.

Be aware that Cisco aggregate-address will not consider a component route unless it is present in the BGP table.

This is done automatically by the feature (you need redistribute or network command for the component route with matching subnet mask with new default no auto-summary an exact match is required). So the check required by your customer may be redundant or unnecessary in my understanding.

Hope to help

Giuseppe

[Giuseppe Larosa]

Hello Frank,

when we use a route-map with network command is to change some attributes of the generated BGP NLRI.

Here you would like to do the same with an aggregate-address command that supports different optional keyworks like suppress-map ,  and so on, where each of them can call a route-map.

I think that what you see can be considered cosmetic if the route-map applied is the same you have configured and provides the desired effects (changing some BGP attributes on the generated aggregate address NLRI).

Check with show ip bpg vrf <vrf-name> <aggregate-address>.

The router is just telling that the invoked route-map will be used to change BGP attributes of the aggregate.

It is the same also in past NX-OS versions see the following example:

https://www.cisco.com/c/m/en_us/techdoc/dc/reference/cli/nxos/commands/bgp/aggregate-address-bgp.html?dtid=osscdc000283

Here route-map is not listed between the available options. (I know it is a different platform and operating system but it is an example)

Hope to help

Giuseppe

[fsebera]

Ok more fun

in reference to the sh bgp vrf <vrf_name> <aggregate>

IOS-XE response

% Command accepted but obsolete, unreleased or unsupported; see documentation.

Nice :-)

Happy Friday

Frank