[BUG] ping 同节点默认网络pod 收到重复回应

[kldancer]

Kube-OVN Version

v1.12.22

Kubernetes Version

v1.31.1

Operation-system/Kernel Version

5.10.0-136.12.0.86.4.hl202.x86_64

Description

ping 同节点默认网络pod 收到重复回应，抓包发现genev_sys_6081 有包。ovs-ovn组件一直刷新对应ip日志

环境信息：

[root@node106 ~]# kubectl get pod -A -owide | grep 110.64.1.41
prometheus-monitoring          prometheus-os-prometheus-0                                   2/2     Running                285 (4h41m ago)   9d      110.64.1.41      node106   <none>           <none>
[root@node106 ~]# ip r
default via 10.210.255.254 dev ens16f0 proto static metric 100 
10.210.0.0/16 dev ens16f0 proto kernel scope link src 10.210.18.106 metric 100 
100.64.0.0/16 dev ovn0 proto kernel scope link src 100.64.0.2 
110.64.0.0/10 via 100.64.0.1 dev ovn0 src 10.210.18.106 
192.166.0.0/16 dev br-ens4f0 proto kernel scope link src 192.166.18.106 
192.167.0.0/16 dev ens6f0 proto kernel scope link src 192.167.18.106 metric 101 
[root@node106 ~]# ip ne 
10.210.18.108 dev ens16f0 lladdr b8:d4:f7:8b:7a:d0 REACHABLE
100.64.0.1 dev ovn0 lladdr 00:00:00:e8:d7:dc REACHABLE
10.210.255.254 dev ens16f0 lladdr 88:df:9e:5b:88:e4 REACHABLE
10.210.18.107 dev ens16f0 lladdr b8:d4:f7:8b:78:8c REACHABLE
10.210.18.30 dev ens16f0 lladdr b8:d4:f7:8b:7a:d0 REACHABLE

Steps To Reproduce

见描述

Current Behavior

ping 同节点默认网络pod 流量不正常

Expected Behavior

ping 同节点默认网络pod 流量正常

[dosubot]

The issue you're experiencing with duplicate ICMP echo replies when pinging a pod on the same node in Kube-OVN version v1.12.22 might be related to the genev_sys_6081 interface. There is a known issue in this version where offloading for the genev_sys_6081 interface was disabled, which could potentially affect network traffic handling ^[1].

To diagnose and potentially resolve this issue, you can use the following steps:

1. Check Logs: Use the kubectl ko log command to gather logs from the kube-ovn components. This can help identify any anomalies or errors related to the network interfaces.

2. Trace and Capture Packets: Utilize the trace or ovn-trace commands to trace the flow of ICMP packets. Additionally, use tcpdump to capture and analyze the network traffic on the ovn0 and genev_sys_6081 interfaces to identify where the duplication might be occurring ^[2].

3. Verify Configuration: Ensure that the configuration of the ovn0 interface is correct. This includes checking the IP address, MAC address, and routes associated with the interface. Misconfigurations here could lead to unexpected network behavior ^[3][4].

4. Network Readiness Check: Perform a network readiness check by pinging the gateway to ensure that the ovn0 interface is operational and correctly configured ^[3].

By following these steps, you should be able to gather more information about the cause of the duplicate ICMP replies and take appropriate actions to resolve the issue.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[oilbeater]

是不是有删除的节点没有清理干净，有遗留的 pod 在旧节点上

[kldancer]

是不是有删除的节点没有清理干净，有遗留的 pod 在旧节点上

是一个系统升级场景，没有删除节点，但是换了个节点作为主节点，待后面复现再观察吧