Infrastructure

The Infrastructure section is where you provision and operate the stateful backing services your applications need — managed databases, message queues, caches, and isolated Kubernetes clusters. In the app it lives at /infrastructure, with tabbed views for Databases, Kafka, and clusters. (The older /cloud URLs now redirect here.)

What you can provision

• Managed PostgreSQL — via CloudNativePG, with backups, a query runner, a schema browser, and per-PR database cloning for preview environments.
• Managed Kafka — brokers and topics for event-driven workloads.
• Managed cache — Redis-compatible caching.
• Isolated clusters (vClusters) — full virtual Kubernetes clusters as a power-user escape hatch, with kubectl, Helm, and CRDs.

Databases, Kafka, and caches are scoped to a service/project; clusters are managed at the organization level.

Working with managed resources

The pfai cloud command set manages these resources from the terminal:

pfai cloud db ...            # managed databases (create, list, query, snapshot)
pfai cloud kafka ...         # managed Kafka brokers and topics
pfai cloud cache ...         # managed cache instances
pfai cloud cluster ...       # virtual clusters
pfai cloud kubeconfig <id>   # fetch a kubeconfig for a cluster
pfai cloud exec ...          # exec into a pod (TTY)
pfai cloud port-forward ...  # forward a local port to a pod or service

The same operations are available over the API under /api/v1/cloud/*, which proxies to the cluster-manager control plane with your identity and permissions enforced.

Direct cluster access

For interactive debugging, proxifai brokers Kubernetes access through the platform (so it works with residential/dynamic-IP clusters and CNPG failover):

• kubeconfig — GET /api/v1/cloud/clusters/{id}/kubeconfig mints a kubeconfig that points back through the platform proxy.
• exec — open a TTY into a pod (WebSocket/SPDY bridge).
• port-forward — forward to a pod, or to a Service that resolves to a ready backing pod at connect time (handles CNPG -rw failover).
• logs follow — stream pod logs.

All of these check the cloud.* permissions and, for mutations, require project membership.

Related