Skip to main content

Service Version And Runtime Readback designed

Platform admins and operators need one place to answer:

What version of each service is running in this environment?

This page defines the target readback surface. It is not a substitute for live cluster evidence; it is the product and ops contract for the page or CLI output that should expose that evidence.

Required Readback Fields

FieldWhy it matters
environmentdistinguishes kind, dev, demo, staging, production
service/component idstable service name such as api, web, billing-worker, node-agent
source committies runtime back to reviewed code
image digest or binary versionproves the deployed artifact, not only the intended commit
config profileshows which environment/profile contract was applied
rollout statusready, degraded, unavailable, unknown
started at / observed atseparates stale readback from current state
health endpoint resultquick runtime health signal
evidence linkdeployment, UAT, or release evidence packet

Component Families

FamilyExamples
Control planeapi, web, terminal-gateway
Workersbilling-worker, provisioning-worker, webhook-worker, notification-relay, outbox-relay
Runtime/fleetnode-agent, provider adapters, terminal/runtime helpers
Data dependenciesPostgres, Redis, NATS, Temporal, object storage
Edge/authCloudflare/Pomerium profile, Keycloak realm/profile
Docs and evidencedocs portal build, Fairway dashboard/runtime

Desired Operator UX

The first implementation can be read-only. Mutating actions such as restart, rollback, promote, or drain should stay behind separately reviewed operator flows.

API / CLI Shape

The page should be backed by an explicit read model rather than direct SQL:

GET /api/.../platform/status/components
gpuaas-cli platform status components --environment kind

The exact route path still needs to follow the route modularization model. The important requirement is that it is API-first and stable enough for UAT, support, and release evidence.

Acceptance Criteria

  • A platform admin can identify stale or mismatched services without reading CI logs.
  • A release operator can prove the deployed commit/image pair after rollout.
  • Security can link a running environment to reviewed release evidence.
  • Product and support can quickly distinguish product bugs from stale deploys.