feat(workflow-executor): add graceful shutdown with in-flight step drain

[matthv]

Summary

• Graceful drain — stop() now waits for in-flight steps to complete before closing resources
• State getter — runner.state exposes the lifecycle: idle → running → draining → stopped
• Configurable timeout — stopTimeoutMs (default 30s) prevents stop() from hanging on stuck steps
• HTTP server stays up during drain — frontend can still query run status while steps finish
• Logger.info — optional method added to Logger interface for drain status messages
• Signal handling is the consumer's responsibility (Runner is a library class)

How it works

SIGTERM received
  → stop() called
    → state = 'draining', polling stopped
    → await in-flight steps (with timeout)
    → close HTTP server, AI client, RunStore
    → state = 'stopped'

Test plan

• State transitions: idle → running → draining → stopped
• State resets to idle on start failure
• stop() waits for in-flight steps before resolving
• stop() resolves after timeout when step is stuck (logs error)
• stop() resolves immediately when no steps in flight
• HTTP server closed after drain, not before
• Drain info/completion logged via Logger.info
• 49 runner tests pass, lint clean

fixes PRD-241

🤖 Generated with Claude Code

Note

Add graceful shutdown with in-flight step drain to workflow executor

• Runner gains a RunnerState (idle → running → draining → stopped) and stop() now waits for all in-flight steps to complete before shutting down, with a configurable stopTimeoutMs (default 30s).
• buildInMemoryExecutor and buildDatabaseExecutor now return a WorkflowExecutor facade with start(), stop(), and state; signal handlers for SIGTERM/SIGINT are registered on start and removed on stop.
• A new unauthenticated GET /health endpoint on ExecutorHttpServer returns 200 when state is running or draining, and 503 otherwise.
• ConsoleLogger and the Logger port gain an optional info() method used to log drain progress.
• Behavioral Change: Runner can no longer be restarted after reaching stopped or draining state, and start() no longer manages an HTTP server directly.

^{Macroscope summarized ba45db9.}

[linear]

PRD-241 Graceful shutdown for the executor

[qltysh]

Coverage Impact

Unable to calculate total coverage change because base branch coverage was not found.

Modified Files with Diff Coverage (4)

Rating	File	% Diff	Uncovered Line #s
	packages/workflow-executor/src/adapters/console-logger.ts	100.0%
	packages/workflow-executor/src/runner.ts	100.0%
	packages/workflow-executor/src/build-workflow-executor.ts	74.4%	97-114
	packages/workflow-executor/src/http/executor-http-server.ts	100.0%
	Total	87.0%

🤖 Increase coverage with AI coding...

In the `feature/prd-241-graceful-shutdown-executor` branch, add test coverage for this new code:

- `packages/workflow-executor/src/build-workflow-executor.ts` -- Line 97-114

🚦 See full report on Qlty Cloud »

🛟 Help

• Diff Coverage: Coverage for added or modified lines of code (excludes deleted files). Learn more.

• Total Coverage: Coverage for the whole repository, calculated as the sum of all File Coverage. Learn more.

• File Coverage: Covered Lines divided by Covered Lines plus Missed Lines. (Excludes non-executable lines including blank lines and comments.)

  • Indirect Changes: Changes to File Coverage for files that were not modified in this PR. Learn more.

[qltysh]

3 new issues

Tool	Category	Rule	Count
qlty	Structure	Function with high complexity (count = 14): createWorkflowExecutor	3