Skip to content

Operations Engine

Status: Draft for review

Purpose

Own cross-cutting runtime operations: asynchronous job execution, scheduling, audit log, observability, operational health, imports/exports, retry/dead-letter tooling and administrative diagnostics.

Owns

  • Job/worker infrastructure contracts
  • Scheduler orchestration
  • Audit records
  • Operational health/status
  • Retry/dead-letter workflows
  • Import/export orchestration
  • Correlation IDs and observability context
  • Operational notifications/incidents metadata

Non-goals

It does not own the business truth being processed by jobs. A job invokes the owning engine rather than mutating another engine's tables directly.

API contract — functional surface

These are contract-level resources and operations, not final controller/file names.

Method Route Primary actor Contract intent
GET /api/platform/v1/operations/health Platform Ops Read platform component health.
GET /api/platform/v1/operations/jobs/failed Platform Ops Inspect failed/dead-letter jobs.
POST /api/platform/v1/operations/jobs/{id}/retry Platform Ops Retry eligible failed job.
GET /api/platform/v1/audit Platform Admin/Support Search platform audit records under authorization.
GET /api/tenant/v1/audit Tenant Owner/Admin Search tenant-visible audit trail.
POST /api/tenant/v1/imports Tenant Admin/Manager Start validated import.
GET /api/tenant/v1/imports/{id} Tenant Admin/Manager Read import validation/progress/result.
POST /api/tenant/v1/exports Tenant Admin/Manager Request approved export.

All mutating operations apply the global tenant, authorization, audit and idempotency rules where relevant.

Events

  • job.failed
  • job.dead_lettered
  • import.completed
  • import.failed
  • export.completed
  • incident.opened
  • incident.resolved

Roles & responsibility matrix

Legend: A accountable, R responsible, C consulted, I informed, none.

Capability Platform Admin Tenant Owner Tenant Admin Specialist/Manager Customer/Visitor
Platform runtime operations A/R I I C
Tenant audit review I A R C
Tenant import/export I A R R
Incident communication A/R I I I I

Dependencies

  • All engines emit auditable facts/jobs/events
  • Platform Engine for authorization/tenant identity
  • Integration Engine for delivery/provider failures

Internationalization Contract

  • Translation assistance/import/export jobs use the same tenant isolation, retries, idempotency, tracing and audit requirements as other async work.
  • Logs/events retain locale identifiers where operationally relevant but should use stable machine error/status codes rather than localized log semantics.
  • Monitoring should expose translation-job/provider failures independently from content publication.

Security / isolation requirements

  • Jobs are tenant-context aware and idempotent.
  • Audit records are append-oriented and access controlled.
  • Operational tooling cannot become an unguarded path to mutate tenant data.

Acceptance criteria

  • Email/order/integration jobs from Laravel have explicit target jobs.
  • Workers can restart without losing durable work.
  • Failed external calls are visible and retryable.
  • Imports validate complete input before unsafe partial mutation where feasible.