0
0
I am using ingress-nginx 0.23.0v in a K8S 1.13v, Calico 3.6v, Kube-Proxy setup.
I am using this as the deployment manifest and this as the service manifest. I have the nodeport and worker IPs added as backends in my hardware LB. But I noticed backend down alerts from my LB frequently and when I tried cURL to the LB backends I can see some of the requests are not getting any result. I took a tcpdump in the NodePort and I can see there is no ACK from the app for these requests which doesn't respond. This is the cURL output for requests that won't get any response.
curl -v 172.17.26.205:30166 * Rebuilt URL to: 172.17.26.205:30166/ * Trying 172.17.26.205... * TCP_NODELAY set
Vishnu Kraj
Posted 2019-03-25T05:33:10.237
Reputation: 11