Administrator guide
This guide gathers the tasks of the platform administrator and points, for each one, to the page that details it. French version: ../administration.md.
Roles and consoles
- The platform administrator holds the
platform_adminrole in the Control Plane. They declare the tenants, activate the licence, read the licence history, the audit and the policy decisions. - Infrastructure operations (hosts, clusters, services, configurations, operations, backups and disaster recovery)
happen in the Platform Manager console, served by
navin-manager, or in the Platform Manager space of the portal, which goes through the Control Plane (ADR-0128,docs/architecture/ADRs/en/ADR-0128-console-unique-platform-manager.md). - Every action takes a policy decision (OPA and OpenFGA) then leaves an audit event, refusals included.
First start
- Install the platform: installation-handbook.md (prerequisites, profiles, offline bundle).
- Register the hosts and create the cluster in the Platform Manager:
navin-manager/docs/en/operations.md(registration tokens, inventory, system settings). - Activate the licence:
POST /v1/license/activationswith the signed token, or the offline command described in licensing.md. The state is read withGET /v1/license. - Connect identity (Keycloak, client federation) and policies: security-hardening.md and
docs/security/en/identity-policy-plane.md. - Configure the components (
EAI_*variables): configuration.md.
Day to day
| Task | Where | Page |
|---|---|---|
| Start, stop, restart a service, rolling restart | Platform Manager, service page | navin-manager/docs/en/operations.md |
| Change a configuration, compare and restore a version | Platform Manager, Configurations | navin-manager/docs/en/operations.md |
| Follow a long operation, abort or retry it | Platform Manager, Operations | navin-manager/docs/en/operations.md |
| Upgrade the platform | offline bundle then services one by one | upgrade.md |
| Back up, restore, exercise disaster recovery | Platform Manager, Backups and DR | ha-dr.md, navin-manager/docs/en/backup-recovery.md |
| Monitor and receive alerts | Prometheus, Alertmanager, Navin Manager (machines and services), portal screens | alerting.md, slo.md |
| Respond to an incident | SOC, Kill Switch | runbooks.md, docs/security/en/soc-runbook.md |
| Renew the licence | renewal token | licensing.md |
| Troubleshoot | logs, states, operations | navin-manager/docs/en/troubleshooting.md |
Rules to know
- A restore needs a plan, a dry run and two approvers other than the requester; the
restore:executepermission is separate frombackup:manage. - A configuration change creates a version; secrets are never shown, only "set" or not.
- The licence limits neither users, nor agents, nor use cases. After the grace period, only new marketplace installations and distribution upgrades are refused; security, audit, backups and running agents never stop.
- The audit is chained and exported to the WORM evidence store: siem-audit.md.
Compliance
The mapping between control families and their evidence is described in compliance-mapping.md.