Skip to content

fix: add skip_exit to drain_listeners to prevent premature pod termination - 2#1062

Closed
santoshpulluri wants to merge 1 commit intorelease/2.0.0-rc1from
spulluri/fix_envoy_drain_issue_v2
Closed

fix: add skip_exit to drain_listeners to prevent premature pod termination - 2#1062
santoshpulluri wants to merge 1 commit intorelease/2.0.0-rc1from
spulluri/fix_envoy_drain_issue_v2

Conversation

@santoshpulluri
Copy link
Copy Markdown
Contributor

Description

When upgrading Envoy from 1.35 to 1.37, pods were being terminated early
during graceful shutdown. The POST /drain_listeners?graceful endpoint causes
Envoy to automatically exit after the drain period (--drain-time-s), which
raced against consul-dataplane's own lifecycle sequencing:

Drain() → wait shutdownGracePeriodSeconds → Quit() → Kill()

Adding &skip_exit decouples the drain phase from the exit decision, ensuring
Envoy stays alive until consul-dataplane explicitly calls Quit() via
/quitquitquit, preserving the intended shutdown lifecycle control.

@santoshpulluri santoshpulluri requested review from a team as code owners April 27, 2026 17:38
@santoshpulluri santoshpulluri changed the base branch from main to release/2.0.0-rc1 April 27, 2026 17:38
@santoshpulluri
Copy link
Copy Markdown
Contributor Author

Wrong Change

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