-
Bug
-
Resolution: Done
-
Medium
-
None
-
None
-
3 node cluster with 2 computes.
Traffic from DC-GW to extraroute connected to VM on compute-0-1 not reachable when vxlan tep is deleted on compute-0-2.
Test Case:
========
->Configure 8 dualstack networks and 16 VMs and 2 routers. Two computes have one VM for each of the 8 networks. Attach 4 dualstack networks to each router
->Configure BGP, 2 VPN instances .
->Attach routers to VPN instances.
->Configure extra routes with four VMs as nexthops for both the routers.
->Verify reachability between DCGW and extraroute.
->Delete tep ip on compute2.
Observation:
Routes of VMs and extraroutes originating from compute2 shoudl be withdrawn -OK
Extraroute should be reachable from DC-GW through compute-0-1-NOK
Issue:
Table 20 entries matching the extra route labels for both ipv4 and ipv6 are removed on compute0-1 due to which the traffic is lost.
RP/0/0/CPU0:ios#
100:24 202f:adc:9213:2f::/64 192.168.113.10 100119 s
100:22 2111:12f:862:18::/64 192.168.113.10 100117 s
root@compute-0-1:~# sh table.sh 20 | grep 100119
root@compute-0-1:~# sh table.sh 20 | grep 100117
100:22 20.0.5.0/24 192.168.113.10 100118 s
100:24 30.0.1.0/24 192.168.113.10 100116 s
root@compute-0-1:~# sh table.sh 20 | grep 100118
root@compute-0-1:~# sh table.sh 20 | grep 100116
root@compute-0-2:~# sh table.sh 20 | grep 100119
root@compute-0-2:~# sh table.sh 20 | grep 100117
Issue description in detail is attached.