A reliable Border Gateway Protocol (BGP) session is vital when your Extreme switch is the hand-off to an ISP or upstream peer. Below is a field-tested checklist—expanded with IPv6 examples—to help you move methodically from physical reachability to policy filters and quickly locate the root cause of missing routes or a stuck session.
Verify the BGP neighbors
Display BGP neighbor information
show bgp neighbor
EXOS # show bgp neighbor
=============================================================================
Peer AS MsgRcvd MsgSent OutQ Up/Down State | PfxRcvd PfxSent PfxAcct
-----------------------------------------------------------------------------------------------------------------------------------
198.51.100.1 64512 38,912 38,840 0 4d03h22m18s Established | 64,820 120 Enabled
2001:db8::1 64513 679 672 0 00h15m12s Established | 1,189 55 Enabled
=============================================================================
Verify that the session is really up
show bgp neighbor 198.51.100.1
Look for the line State : ESTABLISHED. If the FSM is anything else (Idle, Active, Connect, OpenSent/Confirm), focus on layer-3 reachability, TCP/179 ACLs, or authentication first. (Extreme Networks)
[!tip] Finite State Machine (FSM)
FSM stands for Finite State Machine. In BGP, the FSM models how a peering session progresses through six well-defined stages. Each transition is triggered by specific events such as successful TCP handshakes, reception of protocol messages, or timers expiring. Errors push the session back to an earlier state or terminate it.
- Idle – initial state before any attempt to connect
- Connect – TCP port 179 connection attempt in progress
- Active – retrying or awaiting a TCP connection
- OpenSent – BGP OPEN message sent, waiting for peer’s OPEN
- OpenConfirm – OPENs exchanged; waiting for the first Keepalive
- Established – Keepalives and Updates flow; routes are actively exchanged
Knowing the current FSM state helps you pinpoint at which step the peering process is failing, making it a fundamental tool for BGP troubleshooting.
IPv6 peer example
show bgp neighbor 2001:db8::1
The output is identical in structure; you will also see separate capability blocks for ipv6-unicast and ipv6-multicast.
Confirm the negotiated address-families
In the neighbor detail you should see Capabilities Config : ipv4-unicast, … , ipv6-unicast under each address-family section. Missing ipv6-unicast means one side never offered it (often because the right license or knob is absent). Extreme notes that IPv6 BGP requires a Premier/Core license on many Summit platforms. (Extreme Networks)
Inspect what you send to the peer
show bgp neighbor 198.51.100.1 transmitted-routes all
Check that every intended prefix appears and carries the correct Next-Hop, Local-Pref, MED and AS-Path. If something is absent:
- Make sure the network is in the local routing table (RIB).
- Verify any
route-policy out …statements are not filtering it. - Confirm
next-hop-selfis set when advertising iBGP routes to eBGP peers. (Extreme Networks)
EXOS # show bgp neighbor 198.51.100.1 transmitted-routes all
Advertised Routes:
Destination Next-Hop LPref Weight MED AS-Path
----------------------------------------------------------------------------------------
>? 17.17.17.0/24 198.51.100.2 100 0
>? 24.24.24.24/32 198.51.100.21 100 0
>? 198.51.100.0/24 198.51.100.21 100 0
Flags: (*) Preferred BGP route, (>) Active, (d) Suppressed, (h) History
(s) Stale, (m) Multipath, (u) Unfeasible
Origin: (?) Incomplete, (e) EGP, (i) IGP
BGP Route Statistics
Advertised Routes : 3
IPv6 equivalent
show bgp neighbor 2001:db8::1 transmitted-routes ipv6-unicast all
Inspect what you receive
show bgp neighbor 198.51.100.1 received-routes all
EXOS # show bgp neighbor 198.51.100.1 received-routes all
Routes:
Destination Peer Next-Hop LPref Weight MED AS-Path
----------------------------------------------------------------------------------------
*>? 0.0.0.0/0 198.51.100.1 10.58.106.1 100 1 0
*>? 1.1.1.1/32 198.51.100.1 198.51.100.1 100 1 0
*>? 2.2.2.1/32 198.51.100.1 192.168.10.22 100 1 0
*>? 10.58.106.0/24 198.51.100.1 198.51.100.1 100 1 0
*>? 192.168.10.0/24 198.51.100.1 198.51.100.1 100 1 0
*>? 192.168.53.0/24 198.51.100.1 198.51.100.1 100 1 0
*>? 192.168.55.0/24 198.51.100.1 198.51.100.1 100 1 0
? 198.51.100.0/24 198.51.100.1 198.51.100.1 100 1 0
Flags: (*) Preferred BGP route, (>) Active, (d) Suppressed, (h) History
(s) Stale, (m) Multipath, (u) Unfeasible
Origin: (?) Incomplete, (e) EGP, (i) IGP
BGP Route Statistics
Total Rxed Routes : 8
Feasible Routes : 8
Active Routes : 7
Rejected Routes : 0
Unfeasible Routes : 0
Display all routes in the RibIN.
show bgp routes all
For IPv6, simply add ipv6-unicast:
show bgp neighbor 2001:db8::1 received-routes ipv6-unicast all
EXOS # show bgp neighbor 2001:db8::1 ipv6-unicast received-routes all
Routes:
Destination LPref Weight MED
Peer Next-Hop
AS-Path
-------------------------------------------------------------------------
*>i ::/0 100 1 0
2001:db8::1 2001:db8::1
6461
Flags: (*) Preferred BGP route, (>) Active, (d) Suppressed, (h) History
(s) Stale, (m) Multipath, (u) Unfeasible
Origin: (?) Incomplete, (e) EGP, (i) IGP
BGP Route Statistics
Total Rxed Routes : 1
Feasible Routes : 1
Active Routes : 1
Rejected Routes : 0
Unfeasible Routes : 0
Check the local BGP and IP routing tables
For IPv4 BGP
show bgp routes all
show iproute
For IPv6 BGP
show bgp routes ipv6-unicast all
show iproute ipv6
Compare what BGP believes is best with what the FIB (iproute) actually programs. If the best path is missing from the FIB, you may be hitting a hardware route-table limit or an administrative distance conflict.
Typical culprits and quick fixes
| Symptom | Checklist item | Likely fix |
|---|---|---|
| Neighbor stuck in Active | IP reachability / ACLs / auth | enable sharing on LAGs, allow TCP/179, verify passwords |
| No IPv6 routes learned | Capabilities missing | Check Premier/Core license; show licenses or show licenses detail |
| Intended prefix not advertised | Network not in RIB or outbound policy filters it | configure bgp add network …, adjust route-policy out |
| Received route visible but not active | Better path learned elsewhere or policy rejected it | Adjust local-preference, MED, or remove rejecting policy |
| Full-view floods hardware table | Platform route scale limit | Summarize, filter, or switch to partial route-reflector feed |
Handy IPv6 configuration snippets
Advertise a local /48
configure bgp add network ipv6 2001:db8:100::/48
Create an IPv6 eBGP peer
create bgp neighbor 2001:db8::1 remote-AS-number 64512
configure bgp neighbor 2001:db8::1 source-interface ipaddress 2001:db8::20
enable bgp neighbor 2001:db8::1
Display IPv6 routes
show bgp routes ipv6-unicast all
References
How To: How To: Troubleshoot BGP Issues on Switch Engine (EXOS) | Extreme Portal
How To: How to see IPv6 routes transmitted or received by BGP router? | Extreme Portal
How To: How To: Configure BGP on Switch Engine (EXOS) | Extreme Portal
How To: How to add an IPv6 address to a VLAN | Extreme Portal
Q A: Does BGP work with a base license on universal EXOS switches | Extreme Portal
No comments:
Post a Comment