envoyproxy/envoy

allow close downstream idle connection while draining

Open

#42305 opened on Nov 30, 2025

View on GitHub
 (12 comments) (0 reactions) (0 assignees)C++ (27,997 stars) (5,373 forks)batch import
area/drainarea/listenerenhancementhelp wanted

Description

Description: When stopping envoy by using curl -X POST http://localhost:15000/drain_listeners?graceful Envoy only sending Connection: CLOSE on HTTP1, that means idle connection will be Established after idle_timeout at most.

Background: We are building AI Gateway based on Envoy, we hope idle_timeout could be large enough. Meanwhile we hope downstream idle connections could be closed as soon as possible, so that the the Gateway could be shutdown as early as possible.

[optional Relevant Links:] https://www.envoyproxy.io/docs/envoy/latest/operations/admin#operations-admin-interface-drain

Maybe close downstream idle connection could be the default behaviour or add a new configuration to enable it.

Contributor guide

Tech stack
cpp
Domain
backendperformance
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
C++Envoy architectureHTTP connection drainingAdmin interface
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
The issue requests that when draining via POST /drain listeners?graceful, idle downstream connections should be closed immediately instead of waiting for idle timeout. Relevant source files include source/server/drain manager.cc and source/common/http/conn manager impl.cc. Comments discuss options: make it the default or add a new configuration flag. Investigate how idle connections are tracked and where the CLOSE header is set. Review existing tests for drain behavior and the admin handler in source/server/admin/drain handler.cc. A new config option might be introduced in the HCM or listener config.