Console
Sign in
Docs for Database administrator
On this page
  1. What you own
  2. Start here
  3. Common tasks
  4. Reference
  5. Production readiness checklist
  6. When it breaks

Platform operator

You deploy the gateway, Access, Data and the console, keep them healthy and roll them forward or back. This page gathers what you need for that: the order to read things in, the tasks you will repeat, the reference you will look values up in, and where to go when something breaks.

Read the documentation as Platform operator → The sidebar, the home and the previous and next links then follow this page.

What you own#

AreaWhat it covers
DeploymentsFour processes plus gateway-control, their images, replicas, ports and probes
ConfigurationThe GATEWAY_*, ACCESS_*, DATA_* and console variables, the route table, Data's bindings file, manifest and policy bundle
Shared stateRedis or PostgreSQL behind the gateway, Redis behind Access
SecretsGenerating, mounting and rotating them, together with security operations
MonitoringProbes, metrics, alerts, logs
ReleasesUpgrades, rollbacks, tenant cutovers

Start here#

  1. Platform overview and the atlas, for the four processes and what each one talks to.
  2. Quickstart, to see the platform run on one machine.
  3. Platform operations, for the conventions every service shares.
  4. Deploy the platform, for the first production rollout step by step.
  5. Monitor the platform, before you take traffic.

Common tasks#

TaskGuide
First production deploymentDeploy the platform
Bring a new customer onto the platformOnboard a tenant
Publish a new or changed routePut a service behind the gateway, Gateway operations: changing the route table
Wire probes, metrics and alertsMonitor the platform
Upgrade a release, or back it outUpgrade and roll back
Rotate a secretRotate secrets
Move a tenant between deployments of AccessAccess operations: tenant cutover
Sign callers out everywhereGateway operations: revoking sessions
Operate from a terminalCLI usage

Reference#

ServiceConfigurationOperationsTroubleshooting
Gateway and gateway-controlConfigurationOperationsTroubleshooting
AccessConfigurationOperationsTroubleshooting
DataConfigurationOperationsTroubleshooting
ConsoleConfigurationOperationsUser guide
CLISpec: filesUsageUsage: when something refuses
Shared factPage
Ports, addresses and probes of every servicePlatform operations: addresses, probes
How secrets are deliveredPlatform operations: secrets as files
Images, binaries and versionsPlatform operations: images and binaries, versions
The gateway's shared-state rolesGateway operations: shared state
Compose stacks and Kubernetes manifestsDeploy directory

Production readiness checklist#

  • The gateway, Access and Data run in production mode and start cleanly; each validates its environment at start and names what is missing, so a clean start is the first check (platform environment conventions).
  • The gateway and Access use a shared store, not process memory, once more than one replica runs.
  • The control plane's socket stays internal and never sits behind the public ingress (Gateway operations: control plane).
  • Data is reachable only through the gateway: a direct call without the signed identity gets 401.
  • Metrics tokens are set and each /metrics endpoint is scraped.
  • The alerts of each service's operations page are in place.
  • Backups of the Access and Data databases have been restored once in a rehearsal (Back up and restore).
  • The gateway's deployment review is done.

When it breaks#

SymptomStart at
A caller reports an errorTroubleshoot a request
A readiness probe answers 503Gateway, Access, Data
Sign-ins failAccess troubleshooting: failed logins
The gateway answers 503 access_unavailableGateway troubleshooting
A release misbehavesUpgrade and roll back
You suspect a compromiseRespond to an incident