• fix(pegboard): isolate runner config dc lookup failures #4625 : 2 dependent PRs (#4626 , #4630 )
• docs(pegboard): note runner config upsert split write risk #4624
• fix(pegboard): unpack global runner config keys correctly #4623
• fix(pegboard-gateway): enforce tunnel message state #4622
• fix(pegboard-runner): clear terminal tunnel routes #4621 👈 (View in Graphite)
• fix(pegboard): skip protocol version keys in runner pool backfill #4620
• fix(pegboard): create missing normal runner configs #4619
• fix(epoxy): reuse cached optimistic value #4618
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

PR Review: Fix(pegboard-runner): Clear terminal tunnel routes

Overall: This is a clean, correct fix. Tunnel routes are now properly cleaned up after terminal messages, preventing stale entries from lingering in authorized_tunnel_routes.

What's Good

• Correct ordering: Routes are cleared after the message is processed and forwarded. Clearing before would mean a processing failure could permanently lose the route.
• Protocol parity: should_clear_tunnel_route_mk1 and should_clear_tunnel_route_mk2 are symmetric, consistent with the codebase engine runner parity requirement.
• Clean refactor: Extracting route and clear_route before msg is consumed by the publish call is the right approach.
• Test update: Switching the republishes_issued_* tests to use WebSocket messages (which do not clear routes) and asserting the route still exists is a good correctness check.

Issues

Missing tests for the clearing behavior - The PR adds several helper functions (response_start_message_mk2, response_start_message_mk2_with_stream, response_chunk_message_mk2, etc.) but they appear unused. No tests verify that routes are actually removed when terminal messages are sent. Consider adding tests for each clearing case:

• ToServerResponseStart with stream: false -> route gone
• ToServerResponseStart with stream: true -> route still present
• ToServerResponseChunk with finish: true -> route gone
• ToServerResponseChunk with finish: false -> route still present
• ToServerResponseAbort -> route gone
• ToServerWebSocketClose -> route gone

Unused helper functions will generate dead-code warnings - response_start_message_mk2, response_start_message_mk2_with_stream, response_chunk_message_mk2, response_start_message_mk1, response_start_message_mk1_with_stream, and response_chunk_message_mk1 are defined but not used in the tests shown. Either write the tests that use them or remove them.

Minor

The _ => false default in both should_clear_tunnel_route functions is reasonable. A brief comment explaining the intent (non-terminal message kinds keep the route open) would help future protocol additions understand the default behavior.

No Security or Correctness Concerns

The remove_async call on a missing key is a no-op for scc::HashMap, so there is no panic risk if a route was already removed. Route leaks on processing errors are acceptable since an error would propagate and tear down the connection, dropping the entire map.

Preview packages published to npm

Install with:

npm install rivetkit@pr-4621

All packages published as 0.0.0-pr.4621.b2be34f with tag pr-4621.

Engine binary is shipped via @rivetkit/engine-cli on linux-x64-musl, linux-arm64-musl, darwin-x64, and darwin-arm64. Windows users should use the release installer or set RIVET_ENGINE_BINARY.

Docker images:

docker pull rivetdev/engine:slim-b2be34f
docker pull rivetdev/engine:full-b2be34f

npm install rivetkit@pr-4621
npm install @rivetkit/react@pr-4621
npm install @rivetkit/rivetkit-native@pr-4621
npm install @rivetkit/sqlite-wasm@pr-4621
npm install @rivetkit/workflow-engine@pr-4621

Merge activity

• Apr 13, 8:40 PM UTC: A user started a stack merge that includes this pull request via Graphite.
• Apr 13, 8:46 PM UTC: Graphite rebased this pull request as part of a merge.
• Apr 13, 8:47 PM UTC: @MasterPtato merged this pull request with Graphite.