What runs behind the scenes

Three cooperating services implement Loopback; customers usually interact only with the API/UI layer, but understanding roles helps during incidents and security reviews.

API service (service-api)

• HTTP API and authentication/session handling.
• RBAC enforcement using permission keys (see Access control).
• Models persisted in the primary database.
• Enqueue of asynchronous work: system_tasks, message queue messages.

User impact: synchronous validation + “accepted” long-running operations.

Remote execution coordinator

• Executes heavy workflows: workspace creation, Kamaji objects, catalog deployment_create_v2, bundle build/deploy, DNS, secrets, many host/network subroutines.

User impact: minutes-long provisioning pipelines, retries, ordered installs.

Task coordinator

• Scheduled reconciliation for entities (hosts, workspaces, object stores, DNS, monitoring, agents, organizations, scaling groups, …).
• Compute engines talking to Hetzner / IONOS APIs and SSH/kubeadm style host operations.

User impact: continuous alignment — status fields change even when idle; autoscaling decisions.

Data stores & messaging

• Database — system of record.
• Queue / pub-sub — connects API to workers.
• Redis (where used) — ephemeral notifications (e.g. workspace state).

Further reading

• Reconciliation
• Full documentation index