Skip to content

Add NGINX keepalive_timeout flag to ESP startup option #726

@jianhe25

Description

@jianhe25

Google Cloud Load Balancer suggests backend server such as NGINX should change their keepalive_timeout option to 10 minutes to allow load balancer to keep the TCP session alive. See
suggestion https://cloud.google.com/load-balancing/docs/https/#timeouts_and_retries here.

However ESP startup options doesn't include keepalive_timeout option. We have to use a custom NGINX config just to change keepalive_timeout parameter.

Is it possible to add keepalive_timeout to ESP startup options, so we don't have to change it through custom NGINX config?

To be clear, the option is keepalive_timeout

This feature should be a common ask to deploy a scalable service in GCP if customer use both Cloud Endpoints and Cloud Load Balancer.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions