High error rate at DL

[pengnist]

Hi Community,

We are having a few setups running with srsRAN gNB, Open5GS, FlexRIC, and srsue running 1) in separate servers, and 2) with srsRAN gNB, Open5GS, and FlexRIC in the same server. We have tried the connection between gNB and srsue with 1) zmq and 2) USRPs with RF cables and attenuators. In all these deployments, we experience high DL error rate. And after some time, the DL error rate may reach 100%, and then stopped transmission, while the UL still shows traffic, even if application (ping or iperf3) already stopped. I am wondering if anyone has good suggestions how to start the troubleshooting. Could it be the configuration of the system or software?

The servers for the components except srsue have Intel(R) Xeon(R) Gold 6334 CPU @ 3.60GHz.

I attach some screenshots with ZMQ connection and ping 8.8.8.8 -I tun_srsue:

DL stopped:

Thanks

[andrepuschmann]

It's a bit hard to help since you're not providing any logs. Maybe start with one scenario, upload logs and we have a look. But with ZMQ at least you're not supposed to see any BLER of course.

[pengnist]

Please find the attached gnb.log (it has to be compressed to meet size limit). The setup is as follows:

gnb.zip

[pengnist]

Here is the ue.log:
ue.zip

[pengnist]

I am not sure if this info will be useful to find out the cause of the issue, but we are currently using low latency kernel.

[andrepuschmann]

Could you please try again first with the current main release of the gnb and leave all RIC things disabled.

[andrepuschmann]

Is the issue gone?

[pengnist]

Yes, thanks!
The only issue with the new update is that the E2 binder address needs to be specified, if RIC is not in the same machine as the gnb