Hello Juniper Community,
We are currently experimenting with next-table and rib-groups. We have two pair of routing instances running on two MX routers (for redundancy). Instance 1 (which is called internet-global) is used for generating the default route, instance 2 (which is called internet-default) is used for aggregating our public prefixes to instance 1. These instances are currently connected to each other using the lt-interface (logical tunnel). We would like to replace this with next-table and rib-groups, because of the speed limitation. NOTE: instance 1 is a virtual router and instance 2 is a VRF. Changing the routing instance type is not an option for us.
I already made some changes in the config. Here is the config of instance 1:
show routing-instances internet-global
description "EDGE VR, Full table, not in MPLS";
instance-type virtual-router;
interface ae0.10;
interface ae0.101;
interface ae0.211;
interface ae0.4002;
interface ae0.4012;
interface ae0.4013;
interface ae0.4016;
interface lo0.1000;
routing-options {
rib internet-global.inet6.0 {
generate {
route ::/0 discard;
}
}
generate {
route 0.0.0.0/0 discard;
}
autonomous-system xxxx22;
}
Here is the config of instance 2, also a side note, the route are being currently aggregated so I replaced the aggregate with static routes discard for the prefixes and added a policy with the three prefixes exact then accept then reject:
show routing-instances internet-default
description "Internet VRF - axxxx22 subnets only!";
instance-type vrf;
interface ae0.11;
interface ae0.100;
interface ae0.900;
interface ae0.901;
interface ae0.905;
interface ae0.906;
interface ae0.911;
interface ae0.913;
interface ae0.914;
interface ae0.915;
interface ae0.916;
interface ae0.917;
interface ae0.920;
interface ae0.921;
interface ae0.922;
interface ae0.924;
interface ae0.926;
interface ae0.927;
interface ae0.998;
route-distinguisher xxxx76L:100;
vrf-target target:xxxx76L:100;
vrf-table-label;
routing-options {
rib internet-default.inet6.0 {
aggregate {
defaults {
as-path {
path 6xxx9;
}
}
route 2xxx:xxc0::/29;
route 2xxx:xxx0::/29;
}
}
static {
rib-group default-to-global
defaults {
as-path {
path 6xxx9;
}
}
route xxx.xx.32.0/22 discard;
route xxx.xxx.56.0/23 discard;
route xxx.xxx.58.0/24 discard;
route 0.0.0.0/0 next-table internet-global.inet.0
}
router-id xxx.xxx.58.252;
autonomous-system 6xxx9 loops 2 independent-domain;
}
Here is the output of instance 1 routing table:
run show route table internet-global.inet
internet-global.inet.0: 13 destinations, 14 routes (12 active, 0 holddown, 1 hidden)
+ = Active Route, - = Last Active, * = Both
XXX.XX.32.0/22 *[Static/5] 00:07:52
Discard
XXX.XXX.56.0/23 *[Static/5] 00:07:52
Discard
XXX.XXX.58.0/24 *[Static/5] 00:07:52
Discard
XXX.XXX.58.252/32 *[Direct/0] 02:49:55
> via ae0.10
[Local/0] 02:49:55
Local via ae0.10
XXX.XXX.58.255/32 *[Direct/0] 3d 18:48:18
> via lo0.1000
Here is the output of instance 2 routing table:
run show route table internet-default.inet
internet-default.inet.0: 76 destinations, 77 routes (76 active, 0 holddown, 0 hidden)
+ = Active Route, - = Last Active, * = Both
0.0.0.0/0 *[Static/5] 00:09:17
to table internet-global.inet.0
I have a couple of questions about this topic. The first is that the routes are being advertised as discard in the routing table of instance 1, how do I fix this? I want to be able to use them. When I export the prefixes using a policy without the static routes, I don't see them in the routing table and I get no traffic. When I use rib-groups without the import-policy all the routing entries of instance 2 get imported to instance 1 (which we absolutely don't want). Anyone who has experienced this before? One more thing, on instance 1 I have generate routes with discard, do I need to delete those, since I'm using next-table default-route in instance 2?
Thank in advance!
Best regards,
------------------------------
Mohammad Ayash
------------------------------