MPLS L3VPN: Two Sites, One VPN, No Routes
The MPLS core is up and MP-BGP is peering, yet two sites in the same VPN cannot see each other. Find where their routes are dropped on four real routers, right below.
Symptom
LDP is up and MP-BGP is peering, yet two sites in the same VPN cannot see each other. Find where their routes are dropped on four real routers.
Your mission
Prove the cause, then fix it
Format
Four real routers
Need a hint?
Reveal the root cause when you're ready.
Try the investigation first, then use this as your escape hatch, not a dead end.
Create a free account to open MPLS L3VPN: Two Sites, One VPN, No Routes. You get the root cause and the full step-by-step fix.
No spam. The fix lands in your inbox too. Your account includes 5 guided labs free. After that, the Learning Pass unlocks every remaining lab.
Lab debrief
Turn the session into a repeatable troubleshooting pattern.
Use this reference after your attempt: first explain the symptom, then verify the evidence, then confirm the repair.
The problem
A provider runs an MPLS core between two provider edge routers, PE1 and PE2, and connects one customer site to each in VRF CUST. Each site's router, CE1 and CE2, advertises its loopback over eBGP to its PE. The core is up, LDP has exchanged labels and the PEs are peering over MP-BGP, but CE1 has no route to CE2's loopback and CE2 has none to CE1's.
What you'll practice
- Verify an MPLS core: OSPF between the loopbacks, the LDP neighbor and the label bindings
- Check the MP-BGP VPNv4 session between the PE routers
- Follow a customer route from the CE into its PE's VRF and across the core
- Compare each VRF's export and import route-targets and fix the mismatch
- Prove the VPN end to end with routes on both CEs and a labelled traceroute
1. Route distinguisher versus route-target
The route distinguisher is added to each customer prefix to make it unique inside VPNv4, so two customers can use the same addresses. The route-target is a BGP extended community that decides which VRFs import a route. Matching route distinguishers does not make routes import; matching route-targets does.
2. Export and import
A PE tags the routes a VRF sends with that VRF's export route-targets. A receiving PE installs a VPNv4 route into a VRF only when one of the route's route-targets matches one of that VRF's import route-targets.
3. A healthy core can still drop every route
LDP, OSPF and the VPNv4 session can all be up while no customer route crosses the core. When the core checks out, look at the VRF boundary next.
The topology
Four real routers. PE1 and PE2 form the provider core in AS 65000, running OSPF, LDP and an MP-BGP VPNv4 session between their loopbacks. CE1 (AS 65001) attaches to PE1 and CE2 (AS 65002) attaches to PE2, each in VRF CUST and each advertising a loopback over eBGP.
Commands to run yourself
The real diagnostic commands for this lab, copy-paste ready. Run these in the same order to reproduce the investigation on your own session.
show mpls ldp neighborshow bgp vpnv4 unicast all summaryshow ip vrf detail CUSTshow ip route vrf CUSTRate this lab
No ratings yet
Frequently asked
Are these real routers or a simulation?
Real routers. Pressing Start boots four routers just for you, with full command sets: any valid command works, not only the ones the guide suggests. The first start can take a minute or two while the lab server wakes up.
Who can run this lab?
It is included with the Exam Pass and the Year Pass. Each session lasts up to three hours, and you can run one live lab at a time.
Does my exact fix have to match the answer?
No. The lab checks the routers' state: the core sessions, the routes in each PE's VRF, and the routes each CE learns from its PE. Any fix that produces that state passes, and the tempting partial fixes do not.
Ready to run this lab yourself?
No setup, no image sourcing. Book a session or ask for a live demo.