This article details the steps to take to debug an SDWAN spoke device that shows a status of Operational DOWN in Juniper CSO SDWAN.
Device Provisioned status shows UP in CSO.
Device Operational Status shows DOWN in CSO.
Device is processing traffic for customer.
Customer has CLI access apart from CSO UI access, which shows that the device is accessible.
Customer is able to run all commands on the device.
CPU status is normal.
Note: The device in question is an NFX250, which runs in three domains --- WAN(0/1) (JCP <> JDM <> GWR).
JDM hosts the telemetry agent that collects data from GWR and builds a 443-SSH connection towards CSO via the OAM IPsec tunnel, which enables CSO to display operational stats and status.
Steps
It was running in this case.
Data was being written to the agent.log file in this case.
In this case, we saw that the GWR was trying to but was not able to connect to CSO. (The command that helped to see whether GWR was trying to connect to the CSO: show security flow session destination-port 7804 | refresh 2).
We checked routing and saw an ECMP path with two ST tunnels as next-hop.
We inspected another session for the same tunnel interface by using the command show security flow session interface st0.xxxx.
Given this, we concluded that one of the tunnels had two-way traffic for a majority of the sessions and the other tunnel only had one-way traffic.
To allow connection toward the CSO and get it back online, we deactivated the stuck tunnel, which uses the WAN that was in a partially deployed state (command: set interface st0.xxxx disable).
2023-12-11: Article Created