I have created loopback 10.5.40.22 on leaf_2 and 10.5.60.22 on leaf_3 and advertise them in vrf TENANT_1. Both Leaf switches are learning this route via route type 5.
But still it's showing "Network unreachable ' while pinging 10.5.60.22(loopback_Leaf_3) or 10.194.30.25 (host on Leaf_3) source 10.5.40.22
Original Message:
Sent: Aug 04, 2026 06:32 AM
From: DB86
Subject: VXLAN Route reachability from within the switch itself
Worth noting you sourced from the SVI address, not a loopback, so the suggestion above hasn't really been tested yet. Two things to check before you call it a bug.
First, confirm 10.194.10.2 is actually unique to that leaf. Since 10.09.0010 the SVI IP is allowed to be the same as the active gateway IP on every VTEP, and plenty of configs now reuse the same SVI address across leaves. If both leaves answer to .2, it's not a usable source for anything the switch originates itself.
Second, build the loopback properly and advertise it:
interface loopback 2
vrf attach TENANT_1
ip address 10.255.255.1/32
then redistribute connected under vrf TENANT_1, address-family ipv4 unicast, or the far end has no route back and you'll trade the error for silent loss. Then ping source loopback2 vrf TENANT_1. The EVPN VXLAN guide recommends exactly this, a unique loopback as the source, when the SVI carries an anycast gateway.
------------------------------
Dustin Burns
@Worldcom Exchange, Inc.
If my post was useful accept solution and/or give kudos
------------------------------
Original Message:
Sent: Jul 29, 2026 11:40 PM
From: AB-null
Subject: VXLAN Route reachability from within the switch itself
When I ping from the VTEP, it returns "Network is unreachable." The packet is not even exiting the source VTEP. However, the hosts can communicate with each other without any issues.
LEAF-A# sh ip route vrf TENANT_1 | in 10.194.20.
10.194.20.0/24 3.3.3.3 - - B/EV [200/0] 07h:20m:33s
LEAF-A# sh bgp l2vpn evpn route-type 5 | in 10.194.20.
*>i [5]:[0]:[0]:[24]:[10.194.20.0] 3.3.3.3 0 100 0 ?
LEAF-A# ping 10.194.20.22 source 10.194.10.2 vrf TENANT_1
PING 10.194.20.22 (10.194.20.22) from 10.194.10.2 : 100(128) bytes of data.
ping4: sendmsg: Network is unreachable
ping4: sendmsg: Network is unreachable
ping4: sendmsg: Network is unreachable
ping4: sendmsg: Network is unreachable
ping4: sendmsg: Network is unreachable
--- 10.194.20.22 ping statistics ---
5 packets transmitted, 0 received, 100% packet loss, time 4098ms
interface vlan 10
vrf attach TENANT_1
ip address 10.194.10.2/24
active-gateway ip mac 00:00:00:00:00:02
active-gateway ip 10.194.10.254
Host on LEAF_A in vlan 10 can communicate to host on LEAF_B in vlan 20
ITD_SERV_1#sh ip int brief | in 10.194.10.
FastEthernet0/0 10.194.10.23YES manual up up
ITD_SERV_1#ping 10.194.20.22
Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 10.194.20.22, timeout is 2 seconds:
!!!!!
Success rate is 100 percent (5/5),
Original Message:
Sent: Jul 28, 2026 08:53 AM
From: VG-2cb570
Subject: VXLAN Route reachability from within the switch itself
Is the problem on icmp echo packet or the problem on the icmp reply packet ?
Pay attention, that you must source your test traffic (ping, traceroute) from a unicast address, and not from the active-gateway anycast address as the reply
packet might reach the wrong VTEPs. Please configure a unicast loopback in the VRF of interest, source the ping from the loopback IP and report.
Thanks.