Skip to main content

Platform Jobs and Monitoring

Summary​

Platform uses persistence-backed job and provisioning workers, event consumers, an outbox relay, and service-health monitoring. Process liveness and dependency health are distinct signals.

Audience​

  • Platform administrators and support engineers
  • Developers, QA engineers, DevOps engineers, and solution architects

Overview​

Durable job records support leasing, retry, success, and dead-letter outcomes. Provisioning work has a dedicated worker and status model. Messaging components consume supported events and relay persisted outbox messages. Service monitoring probes registered health targets and records current and historical results.

Purpose​

These components make asynchronous control-plane work observable and recoverable without treating a basic process-health response as proof that every dependency is functioning.

Architecture​

Business concepts​

  • Durable platform jobs — Implemented foundation: records move through pending, running, succeeded, and dead-lettered outcomes with lease and retry metadata.
  • Provisioning worker — Foundation: an optional background worker claims eligible isolated-database provisioning requests and reports completion or failure.
  • Background processing — Implemented: hosted services are registered for supported workers, monitoring, consumption, and relay functions.
  • Event consumption — Implemented foundation: Platform consumes selected messages and records processed-event/audit state.
  • Outbox relay — Implemented foundation: messages are persisted before publication and relayed asynchronously.
  • Service monitoring — Implemented: registered targets can be classified as healthy, degraded, offline, or unknown, with response and heartbeat history.
  • Operational adoption — Requires confirmation: enabled workers, schedules, alert routing, service registry targets, and production runbooks are environment-owned decisions.

Technical implementation​

The durable queue uses atomic persistence operations to claim available work and prevent concurrent ownership. Internal schedules and queue mechanics are intentionally omitted. Monitoring performs configured service probes; a successful Platform process-health response confirms the responding process, while dependency-aware service checks provide a broader—but still bounded—view.

Source References​

  • microservices/src/platform-service/Program.cs
  • microservices/src/platform-service/Jobs/PlatformDurableJobQueue.cs
  • microservices/src/platform-service/Jobs/JobRecord.cs
  • microservices/src/platform-service/Provisioning/TenantProvisioningWorker.cs
  • microservices/src/platform-service/Messaging/OutboxRelayHostedService.cs
  • microservices/src/platform-service/Messaging/PlatformEventConsumer.cs
  • microservices/src/platform-service/Messaging/ServiceHealthMonitor.cs
  • microservices/src/platform-service/Application/ServiceRegistryService.cs
  • microservices/src/platform-service/Api/PlatformJobsEndpoints.cs
  • microservices/src/platform-service/Api/PlatformServicesEndpoints.cs

See Also​

Keywords​

  • Durable processing
  • Outbox relay
  • Service probing
  • Process health

Revision Information​

  • Status: Draft
  • Last reviewed: 2026-07-14
  • Review cycle: Quarterly