How to change Kubernetes NodePort service timeout?
I have tested it by running infinite loop php code using apache2 webserver but after 3 minutes it ended up with 504 gateway timeout from nginx. I can guess that it handled by NodePort service Kubernetes
Most of the service aspects in Kubernetes are handled by the kube-proxy and/or regular iptables, so there is not a TCP connection timeout per se. If you are looking at lower-level primitives you could look at TCP timeouts at the Linux kernel level ⏳, but typically those have defaults that are reasonable.
You provided very few details on your question but assuming you are using an Nginx ingress controller, you can weather the services/pods that actually service traffic are actually up and receiving traffic at the specific moment that you are getting those timeouts. You can also tweak the timeouts ⏳ at the Nginx ingress controller level using annotations or the configmap
✌️
NodePort simply exposes pod's ports on the node, health checking is not its functionality.
I guess you are looking for livenessProbe that can be defined per container. Its purpose is to check whether a container is healthy and if that is not the case, it will kill it and then restart is so that the container is returned to its original/normal state.
Example of HTTP liveness probe that checks health of a container every 5 seconds by hitting port 80 and path / of the container.
livenessProbe:
httpGet:
port: 80
path: /
periodSeconds: 5
more configuration detail here