fix: emit terminal event for operation failures

[tomdps]

Summary

• emit one current-run CLUSTER_FAILED event when a cluster operation chain fails
• let the existing terminal subscription own cluster shutdown
• cover both operation-validation failure paths with exact-one-terminal assertions

Root cause

CLUSTER_OPERATIONS failures emitted only the nonterminal CLUSTER_OPERATIONS_FAILED diagnostic before stopping directly. Foreground result writers therefore saw no authoritative terminal event and rejected the retained result.

Validation

• focused integration tests: 2 passing
• full test suite: 2,901 passing, 18 pending
• typecheck and lint (no errors)
• Opcore and Opcore Zero clean
• independent review: approved with no P0-P3 findings

[greptile-apps]

Greptile Summary

The PR terminalizes failed cluster-operation chains with one current-run CLUSTER_FAILED event and delegates shutdown to the existing terminal subscription.

• Scopes terminal-event deduplication to the current cluster run.
• Removes direct shutdown from the operation-failure handler.
• Adds integration assertions for exactly one terminal failure event across both validation-failure paths.

Confidence Score: 5/5

The PR appears safe to merge with operation failures now producing the terminal event required by result consumers while preserving shutdown behavior.

Current-run boundaries exclude historical terminal records, synchronous message delivery reaches the pre-registered shutdown subscriber, and the updated tests cover both validation-failure paths.

Important Files Changed

Filename	Overview
src/orchestrator.js	Adds current-run terminal-event guarding and routes operation failures through the established terminal shutdown subscription without an accepted defect.
tests/integration/orchestrator-flow.test.js	Extends both operation-validation failure scenarios with exact-one CLUSTER_FAILED assertions.

Sequence Diagram

sequenceDiagram
  participant Op as CLUSTER_OPERATIONS handler
  participant Bus as MessageBus / Ledger
  participant Terminal as CLUSTER_FAILED subscription
  participant Cluster as Orchestrator.stop
  Op->>Op: Operation chain rejects
  Op->>Bus: Publish CLUSTER_OPERATIONS_FAILED
  Op->>Bus: Query current-run terminal events
  alt No terminal event exists
    Op->>Bus: Publish CLUSTER_FAILED
    Bus->>Terminal: Deliver terminal event synchronously
    Terminal->>Cluster: stop(clusterId)
  else Current run already terminal
    Op-->>Op: Suppress duplicate terminal event
  end

Loading

_{Reviews (1): Last reviewed commit: "fix: emit terminal event for operation f..." | Re-trigger Greptile}

[github-actions]

🎉 This PR is included in version 6.39.0 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀