Hi,
went through archive but can't find anyone facing the same problems.
We have a set of 5 CER (3 of them RT running bgp full table). Ospf/Mpls/IBGP4vpnve full mesh,EBGP running.
We have a large l3vpn (vrf) running bgp multipathing and ebgp as PE-CE Routing Protocol. Two of them inserting a default route into vrf.
We started seeing issue when upgrading to 5.4a (at that time not running any RT's). If a route in this VRF flaps one of the CER's starts blackholing traffic. It is announcing the prefixes to peers but not forwarding the traffic. At first view labels look correct.
Running "clearip bgp vrf <VRF-NAME> neigh all" clears the error.
Starting with the "new" -RT's running firmware 5.6d we can force this issue by
1. Creating a new vrf
2. Create bgp address family for this VRF
When at point 2 the big VRF breakes on this router.
When it was broken we did run the following commands on all routers to fetch debug info before clearing the bgp neighbors:
show tech
show ip bgp vpnv4 neighbors <peer>.74 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.74 routes det
show ip bgp vpnv4 neighbors <peer>.74 routes-summary
show ip bgp vpnv4 neighbors <peer>.75 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.75 routes det
show ip bgp vpnv4 neighbors <peer>.75 routes-summary
show ip bgp vpnv4 neighbors <peer>.69 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.69 routes det
show ip bgp vpnv4 neighbors <peer>.69 routes summary
show ip bgp vpnv4 neighbors <peer>.84 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.84 routes det
show ip bgp vpnv4 neighbors <peer>.84 routes-summary
sh mpls forwarding
sh ip bgp vrf <VRF> routes det
sh ip bgp vrf <VRF> tags
any hints on additional commands or on what to look for in the debug text files?
Regards,
Nils
went through archive but can't find anyone facing the same problems.
We have a set of 5 CER (3 of them RT running bgp full table). Ospf/Mpls/IBGP4vpnve full mesh,EBGP running.
We have a large l3vpn (vrf) running bgp multipathing and ebgp as PE-CE Routing Protocol. Two of them inserting a default route into vrf.
We started seeing issue when upgrading to 5.4a (at that time not running any RT's). If a route in this VRF flaps one of the CER's starts blackholing traffic. It is announcing the prefixes to peers but not forwarding the traffic. At first view labels look correct.
Running "clearip bgp vrf <VRF-NAME> neigh all" clears the error.
Starting with the "new" -RT's running firmware 5.6d we can force this issue by
1. Creating a new vrf
2. Create bgp address family for this VRF
When at point 2 the big VRF breakes on this router.
When it was broken we did run the following commands on all routers to fetch debug info before clearing the bgp neighbors:
show tech
show ip bgp vpnv4 neighbors <peer>.74 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.74 routes det
show ip bgp vpnv4 neighbors <peer>.74 routes-summary
show ip bgp vpnv4 neighbors <peer>.75 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.75 routes det
show ip bgp vpnv4 neighbors <peer>.75 routes-summary
show ip bgp vpnv4 neighbors <peer>.69 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.69 routes det
show ip bgp vpnv4 neighbors <peer>.69 routes summary
show ip bgp vpnv4 neighbors <peer>.84 advertised-routes det
show ip bgp vpnv4 neighbors <peer>.84 routes det
show ip bgp vpnv4 neighbors <peer>.84 routes-summary
sh mpls forwarding
sh ip bgp vrf <VRF> routes det
sh ip bgp vrf <VRF> tags
any hints on additional commands or on what to look for in the debug text files?
Regards,
Nils