How to configure proxy_connect_timeout for TCP services in ingress-nginx v1.12.1?

[carefree666]

Hi team,

I'm using ingress-nginx v1.12.1 and have both HTTP and TCP services configured.

For HTTP ingress, we can easily control the connection timeout via the ConfigMap key:

proxy-connect-timeout: "5s"

However, for TCP services (configured through the flag --tcp-services-configmap=$(POD_NAMESPACE)/tcp-services), I cannot find a documented way to set the equivalent stream-level directive proxy_connect_timeout
.

For example, the generated stream configuration looks like this:

server {
listen 25001;
proxy_timeout 600s;
proxy_next_upstream on;
proxy_next_upstream_timeout 600s;
proxy_next_upstream_tries 3;
proxy_pass upstream_balancer;
}

But there’s no proxy_connect_timeout applied by default, which causes long connection stalls when the upstream pod is down (resulting in errors like (111: Connection refused) or (110: Operation timed out)).

I tried setting values like this in the main nginx-configuration ConfigMap:

proxy-connect-timeout: "5s"

…but it only affects the HTTP layer, not the stream (TCP) layer.

My question:

Is there an officially supported way to configure proxy_connect_timeout for TCP/stream services when using --tcp-services-configmap?
If not, is the recommended approach to use the global stream-snippet field in the ConfigMap, or will there be future support for per-service stream configuration?

Thanks!

[k8s-ci-robot]

This issue is currently awaiting triage.

If Ingress contributors determines this is a relevant issue, they will accept it by applying the triage/accepted label and provide further guidance.

The triage/accepted label can be added by org members by writing /triage accepted in a comment.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.