K8s cluster blocks udp traffic reach to upstream server when source IP getting changed(to support source ip preservation)

Asked Sep 23 '21 at 00:09

Active Sep 24 '21 at 22:42

Viewed 179 times

As soon as I add the ‘use_orginal_src_ip’ configuration in envoy proxy to preserve the source IP, the upstream server stops receiving traffic.

Setup details: network device outside k8s cluster -> envoy proxy(k8s pod) -> upstream service(k8s pod)

TCPDUMP captures UDP traffic on the k8s host, sent by proxy to the upstream server. However, the UDP packet getting dropped as it never reaches the upstream server.

edited Sep 24 '21 at 22:42

asked Sep 23 '21 at 00:09

Rajen Patel

Which version of Kubernetes did you use and how did you set up the cluster? What version of envoyproxy did you use? – Mikołaj Głodziak Sep 23 '21 at 12:50
envoy: v1.19.1 Kubernetes : v1.17.1 deployed using kudeadm and calico cni. It seems k8s calico detects IP spoofing and drops the packets(envoy proxy does it job well). I have disabled reserve path forwarding as well. – Rajen Patel Sep 24 '21 at 20:47
Is this issue fixed now? – Sounak Saha Feb 04 '22 at 10:17

K8s cluster blocks udp traffic reach to upstream server when source IP getting changed(to support source ip preservation)

0 Answers0