Unable to gracefully terminate NETCONF Call Home session from server side (ORAN ORU setup)

[mradul-29]

Hi @michalvasko
We are working on an O-RAN based ORU (O-RU) implementation using Netopeer2 + sysrepo (libnetconf2 underneath) on an embedded Linux platform.

Environment:
NETCONF server: Netopeer2-server (running as: netopeer2-server -v3 -t 10)
libnetconf2 v3.5.4
Backend: sysrepo v2.2.36
Transport: SSH (Call Home enabled)
Architecture: ORU and ORU-Controller
Connection type: persistent Call Home
Platform: embedded Linux (watchdog enabled, process restart sensitive) (aarch64)

Current Behavior :
We successfully establish a NETCONF Call Home connection from the ORU (server side) to the controller.
However, we are facing an issue when trying to terminate the active Call Home session from the ORU side.

What We Tried:

• Sysrepo config change
  Removed call-home block from ietf-netconf-server running sysrepo
  Result: prevents future reconnections, but existing session remains active
• Signals to netopeer2-server
  SIGUSR1 → session closes, but watchdog connection drop → resulted in system reboot
  SIGTERM → causes full service restart and system rebooted
• Network-level methods
  iptables / socket-level attempts not viable in this environment
• Kill-session
  Attempted to trigger the kill-session RPC from using the Sysrepo sr_rpc_send_tree API.
  The implementation looks like this:

The sr_rpc_send_tree call returns an error.

We need a way to: Gracefully terminate an active Call Home NETCONF session from the ORU side, without:

• restarting netopeer2-server
• triggering watchdog
• relying on controller-side disconnect

Question:
1.) Is there a supported way in Netopeer2/libnetconf2 to cleanly close an active Call Home session from the server side (without restarting the process)?
2.) Is controller-side disconnect the only expected approach for this scenario?

Thanks!

[michalvasko]

We need a way to: Gracefully terminate an active Call Home NETCONF session from the ORU side

In NETCONF it is not possible to terminate the session gracefully from the server side so I cannot help you.

1. You can connect to the server with another client and kill the other session. Not exactly clean and requires another NETCONF session but it is the only way.
2. Yes, that is what is expected in normal circumstances.

[mradul-29]

Okay, Thanks!

[mradul-29]

Hi @michalvasko

As you suggested, I tried using a dummy client and then killing the existing one.
For this, I made a few changes to the command used to start the Netopeer server so that we can have the netopeer2-server.sock file.
Earlier, we were using: netopeer2-server -v3 -t 10 , Now, I have changed it to: netopeer2-server -v3 -t 10 -U /var/run/netopeer2-server.sock
However, with the updated command, sometimes the Netopeer server is not starting, and we are also unable to see any failure logs.

Could you please help check what might be causing this?

[michalvasko]

However, with the updated command, sometimes the Netopeer server is not starting, and we are also unable to see any failure logs.

I am really not able to help with such a generic problem description.