UPF Cli Run Error

jialec1909 · November 25, 2022, 4:36pm

Hello,

I have trouble when running the test ./run.sh

And I looked up the IP routing table, gtp5g has been removed. Could you help me with that, please ?

Best regards

meow0122 · February 15, 2023, 6:27am

Hi @jialec1909 ,
Regarding your problem, you can refer to this issue

github.com/free5gc/free5gc

[Bugs] UPF Cli Run Error: open Gtp5g: open link: create: file exists

opened 02:14AM - 03 Jan 23 UTC

closed 03:30AM - 29 Jun 23 UTC

tywofxd

**We will advise people to follow the issue template set, otherwise the issue mi…ght be disregarded.** **free5GC mainly uses GitHub for issue tracking. Information regarding to general questions or technical support. It will be highly considered if forward to the [forum](https://forum.free5gc.org).** **free5GC will appreciate it, if people can refer to [TS](https://github.com/free5gc/free5gc/wiki/Trouble_Shooting) and [forum](https://forum.free5gc.org) prior to bug reporting** ## Describe the bug I use UERANSIM to open multiple UE threads to interact with the 5G core. The python script for opening such UE threads is like the following: os.system("nohup " + UERANSIM_build_dir + "nr-ue -c ./config/%s &" % file) Then I use UERANSIM's nr-cli to make the UE thread to interact with the 5g core: os.system(UERANSIM_build_dir +"nr-cli -e '%s' %s" % (command, ueId)) However, when the number of UE thread is up to 50, the UPF pod crashes, throwing the following error. 2023-01-03T00:36:39Z [ERRO][UPF][Main] UPF Cli Run Error: open Gtp5g: open link: create: file exists Less UE threads will not lead to this error. ## To Reproduce Steps to reproduce the behavior: 1. Use this helm repo to deploy the 5g core and 5g RAN on a K8s cluster: https://github.com/Orange-OpenSource/towards5gs-helm 2. kubectl exec -it into the ue pod, and use this repository (https://github.com/tywofxd/5GLogGenerator) to generate multiple UE threads to interact with the 5g core. 3. run command: python3 run.py --ue 50 --sec 36000 4. UPF crashes after several minutes. ## Expected behavior I want the 5g core to operate normally with many UE threads interacting with it. ## Screenshots ![image](https://user-images.githubusercontent.com/39151745/210288781-d53cb9f3-15eb-446d-8a92-bea927230237.png) ## Environment (please complete the following information): - free5GC Version: v3.2.0 - OS: Ubuntu 18.04 Server - Kernel version: 5.4.0-131-generic - helm chart: https://github.com/Orange-OpenSource/towards5gs-helm - k8s version: v1.23.4 - go version: [e.g. 1.10.0 linux/amd64] - c compiler version (Option): [e.g. gcc version 1.1.0] ## Trace File ### Configuration File [free5gc-charts.zip](https://github.com/free5gc/free5gc/files/10334033/free5gc-charts.zip) ### PCAP File Dump the packet and provide the pcap file here. If not clear of what to do, this command can be used `sudo tcpdump -i any -w free5gc.pcap` prior to running bug reproduce. Then upload the pcap file `free5gc.pcap`. ### Log File ![image](https://user-images.githubusercontent.com/39151745/210292174-19e6c89b-8a7a-467a-990b-2e14ffa7ae10.png) ## System architecture (Option) Please provide the draft architecture, including the scenario, use cases, installation environment(bare metal, vm, container, or k8s), etc. ## Walkthrough (Option) free5GC will be interested on the research or finding in brief. ## Additional context My repository for generating UE threads is here: https://github.com/tywofxd/5GLogGenerator

You can use ./force_kill.sh or sudo ip link del upfgtp to recover it when the error happened.