A systematic workflow separates transport/session problems from route advertisement, policy, best-path, RIB and forwarding problems.
Learning objectives
- Interpret summary and FSM states.
- Troubleshoot TCP/179 and neighbour configuration.
- Diagnose missing routes, withdrawals and policy issues.
- Check next-hop reachability and best-path selection.
- Separate BGP table, RIB and forwarding outcomes.
- Use debug and packet capture safely.
1. Layered workflow
- Check neighbour state and record exact FSM state.
- Verify peer address, source interface/address and IP reachability.
- Check TCP/179, ACL/firewall and authentication.
- Verify local ASN, remote-as, update-source and eBGP multihop.
- Inspect neighbour detail, notifications and counters.
- Check route origination, UPDATE/withdrawal and address-family activation.
- Inspect inbound/outbound policy and received/advertised routes.
- Check next-hop reachability and best-path selection.
- Check BGP table, RIB installation and forwarding separately.
- After correction, measure convergence and retain evidence.
2. Key commands
show ip bgp summary
show ip bgp neighbors
show ip bgp <prefix>
show ip route <prefix>
show ip route <peer-address>
show tcp brief
show logging
show ip prefix-list
show route-map
received-routes may require inbound soft-reconfiguration and platform support. Command availability and address-family syntax vary.
3. FSM symptom guide
| State / symptom | Checks |
|---|---|
| Idle | Administrative shutdown, configuration, resource or policy conditions; inspect logs/config. |
| Connect | TCP attempt in progress; verify reachability and TCP/179. |
| Active | TCP connection retries; not Established. Check routing, ACLs, source address and peer listener. |
| OpenSent | TCP connected and OPEN sent; inspect ASN, capabilities and OPEN validation. |
| OpenConfirm | Waiting for KEEPALIVE/protocol progress; inspect timers, capture and notifications. |
| Established but no prefixes | Check origination, address family, filters, route maps, refresh and peer advertisement. |
4. Missing route investigation
- Confirm source route exists in originating router RIB.
- For Cisco IOS network statements, confirm exact prefix/mask.
- Check redistribution and route-map/prefix-list policy.
- Inspect advertised routes at sender and accepted routes at receiver.
- Check AS_PATH loop prevention and next-hop reachability.
- Inspect best path and RIB installation.
- Verify forwarding entry and actual traffic.
5. UPDATE, withdrawal and capture
An UPDATE can advertise reachable prefixes and/or withdraw previously advertised prefixes. Determine whether the source withdrew the route, policy filtered it, the session reset or best-path/RIB selection changed.
Wireshark: tcp.port == 179
Linux:
ping <peer-address>
traceroute <peer-address>
nc -vz <peer-address> 179
ICMP failure alone does not prove BGP is unreachable because ICMP may be filtered. Debug commands can be CPU-intensive; use cautiously, preferably in a lab or controlled window.
6. Root-cause report
| Field | Record |
|---|---|
| Issue / symptom | What failed and when |
| Topology / software | Router IDs, ASNs, interfaces, image/version |
| Expected behaviour | What should have happened |
| Actual behaviour | Observed FSM, route and forwarding state |
| Evidence | CLI, logs, packet capture and timestamps |
| Root cause | Confirmed cause, not symptom |
| Correction | Configuration/operational change |
| Regression | Tests repeated and results |
| Impact / recovery | Route loss, packet loss and convergence |
Senior QA scenarios
| ID | Test | Expected result |
|---|---|---|
| N-01 | Block TCP/179 | Session cannot establish; transport issue diagnosed. |
| N-02 | Wrong remote-AS | OPEN validation fails; notification/log inspected. |
| N-03 | Shutdown peering interface | Session tears down and routes withdraw. |
| N-04 | Restore connectivity | Session and routes recover. |
| N-05 | Established but filter all prefixes | Session up; route table reflects policy. |
| N-06 | Withdraw advertised prefix | Withdrawal propagates. |
| N-07 | Unreachable next hop | Route unusable/not installed as expected. |
| N-08 | Wrong update-source/multihop | Session failure diagnosed. |
| N-09 | Authentication mismatch | Establishment fails. |
| N-10 | Controlled Hold expiry | Teardown/log/capture align. |
| N-11 | Compare BGP table and RIB | Installation discrepancy explained. |
| N-12 | Check forwarding plane | Forwarding matches installed route. |
Interview questions
- What does Active mean?
- First check when peer stuck Connect/Active?
- Can session be Established with no routes?
- How troubleshoot missing prefix?
- Withdrawal versus filtered route?
- Why check next-hop reachability?
- BGP table vs RIB vs FIB?
- Why can ping failure be inconclusive?
- When use Wireshark?
- Why use debug cautiously?
- How test recovery?
- What belongs in a root-cause report?
Student Knowledge Assessment – Module N (reconstructed)
Q1. Which command gives a quick overview of BGP peer state?
A. show ip bgp summary
B. show vlan
C. show arp
D. show clock
Correct answer: A. show ip bgp summary
Explanation: The summary displays neighbour states and prefix counts.
Q2. What does BGP Active state indicate?
A. Session fully operational
B. Router attempting/retrying TCP establishment
C. Route is best
D. Peer administratively up
Correct answer: B. Router attempting/retrying TCP establishment
Explanation: Active is a TCP establishment/retry state, not Established.
Q3. What is a first transport check for Connect/Active?
A. TCP/179 reachability
B. DNS MX
C. NTP stratum
D. VLAN name
Correct answer: A. TCP/179 reachability
Explanation: Check IP reachability and TCP port 179.
Q4. Can a BGP session be Established with zero prefixes received?
A. No
B. Yes
C. Only OSPF
D. Only route reflectors
Correct answer: B. Yes
Explanation: Session can be Established while origination or policy prevents route exchange.
Q5. What does an UPDATE carry?
A. Only timers
B. Route advertisements and/or withdrawals
C. Only TCP ACKs
D. Only router IDs
Correct answer: B. Route advertisements and/or withdrawals
Explanation: UPDATE advertises reachable prefixes and can withdraw routes.
Q6. What should be checked when a prefix is missing?
A. Origination, policy, next hop and best path
B. Only hostname
C. Only interface description
D. Only NTP
Correct answer: A. Origination, policy, next hop and best path
Explanation: Check the complete advertisement and selection path.
Q7. Does ping failure prove TCP/179 unavailable?
A. Yes
B. No
C. Only eBGP
D. Only iBGP
Correct answer: B. No
Explanation: ICMP may be filtered; test TCP and inspect routing/firewall evidence.
Q8. What does next-hop reachability affect?
A. Whether a path can be usable/installed
B. Only hostname
C. Only TCP port
D. Only community display
Correct answer: A. Whether a path can be usable/installed
Explanation: An unreachable next hop can prevent route usability or RIB installation.
Q9. Why distinguish BGP table, RIB and FIB?
A. Different stages of selection and forwarding
B. They are identical
C. Only IPv6
D. Only MED
Correct answer: A. Different stages of selection and forwarding
Explanation: A route can be in BGP but not installed in RIB/FIB.
Q10. Why use debug cautiously?
A. It can consume CPU/resources and generate large output
B. It changes ASN
C. It encrypts traffic
D. It disables route maps
Correct answer: A. It can consume CPU/resources and generate large output
Explanation: Debugging can affect router performance.