This is an investigation story where all pods on one host lost connectivity to external service. It looks like something below: $ nsenter --net=/proc/29080/ns/net curl docs.cilium.io -v > GET / HTTP/1.1 > Host: docs.cilium.io > User-Agent: curl/7.79.1 > Accept: */* > Here 29080 is the pid of some pod process, and it appeared that TCP connection has been established, but somehow it couldn’t receive any response. About the cluster The cluster was set up like something below:
