Skip to content

tcp_proxy: remove tcp_proxy_set_idle_timer_immediately_on_new_connection runtime guard#44076

Open
wdauchy wants to merge 1 commit intoenvoyproxy:mainfrom
wdauchy:remove-tcp-proxy-idle-timer-runtime-guard
Open

tcp_proxy: remove tcp_proxy_set_idle_timer_immediately_on_new_connection runtime guard#44076
wdauchy wants to merge 1 commit intoenvoyproxy:mainfrom
wdauchy:remove-tcp-proxy-idle-timer-runtime-guard

Conversation

@wdauchy
Copy link
Contributor

@wdauchy wdauchy commented Mar 22, 2026

Commit Message:
This flag was added in 1.36.0 and has been defaulting to true for multiple releases. The idle timer is now always started immediately on new connections.
Additional Description:
Risk Level:
Testing:
Docs Changes:
Release Notes:
Platform Specific Features:
[Optional Runtime guard:]
[Optional Fixes #Issue]
[Optional Fixes commit #PR or SHA]
[Optional Deprecated:]
[Optional API Considerations:]

…ion runtime guard

This flag was added in 1.36.0 and has been defaulting to true for
multiple releases. The idle timer is now always started immediately
on new connections.

Signed-off-by: William Dauchy <william.dauchy@datadoghq.com>
@repokitteh-read-only
Copy link

As a reminder, PRs marked as draft will not be automatically assigned reviewers,
or be handled by maintainer-oncall triage.

Please mark your PR as ready when you want it to be reviewed!

🐱

Caused by: #44076 was opened by wdauchy.

see: more, trace.

@wdauchy wdauchy marked this pull request as ready for review March 22, 2026 20:33
@repokitteh-read-only
Copy link

CC @envoyproxy/runtime-guard-changes: FYI only for changes made to (source/common/runtime/runtime_features.cc).

🐱

Caused by: #44076 was ready_for_review by wdauchy.

see: more, trace.

@wdauchy
Copy link
Contributor Author

wdauchy commented Mar 23, 2026

Removing tcp_proxy_set_idle_timer_immediately_on_new_connection folds everyone onto the immediate idle-timer path; changelog entry documents the dropped runtime knob. nit: anyone who forced the guard off loses that escape hatch—release notes are the only migration signal.

Thanks for the review! The previous behavior (not starting the idle timer immediately) was an actual bug that caused connection leaks when the upstream connection wasn't fully established. In my own opinion, there's no legitimate reason to keep the old code path as disabling the guard would mean opting back into a known resource leak(?).

This follows the standard Envoy process for runtime guard removal: the fix shipped in 1.36.0, has been defaulting to true for 2 releases, and the changelog entry documents the removal for anyone tracking runtime flag changes.

do you have other concerns?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant