25. Managed infrastructure
The hub for the heavier things Daalu runs near your environment — federated clusters, observability, and the servers and switches it manages for you.
At a glance
| What it is | A tabbed hub for the infrastructure Daalu operates close to you: federated Kubernetes clusters and their observability, the network-and-server management surface, and the onboarding wizard that stands it all up. |
| Where to find it | https://ops.daalu.io/managed-infra |
| Who can use it | Everyone can view; connecting clusters and provisioning management stacks is admin-only. |
Managed infra is where the parts of Daalu that live inside or next to your environment are set up and watched. It’s distinct from Integrations (Chapter 24), which is about connecting external SaaS and pulling data in. Think of it this way: Integrations is how Daalu sees your world; Managed infra is the infrastructure Daalu runs in it.
The page is a tabbed hub with three tabs.
Tab 1 — Clusters & observability
This tab is the inventory of your federated Kubernetes clusters and the observability stacks that feed the Assistant.
Federated clusters
Each row is one cluster connected through Daalu Edge — the outbound-only tunnel that lets the Daalu cloud reach services behind your firewall without you opening a single inbound port. The concept is Chapter 15; the deployment walkthrough is Chapter 41.
A row shows the cluster’s name, tunnel health, and time since the last handshake. Click it to open:
- Overview — node and pod counts, top-level health from the API server.
- Nodes — status, capacity, and labels per node.
- Recent rollouts — Deployments, StatefulSets, and DaemonSets that changed in the last 24 hours.
- Reconciliation — drift between cluster manifests and live state, if enabled.
- Tunnel diagnostics — peer info, last handshake, and bytes in and out.
The Add cluster button generates a one-shot bootstrap invite. You install the Daalu Edge chart in your cluster with it, the tunnel comes up, and the row turns connected within about thirty seconds.
Why it matters — A connected cluster is what makes the Assistant’s kube tools, the GPU router, and your own-GPU inference work. Nothing in your cluster is exposed to the public internet; revocation is instant — delete the edge chart and the tunnel dies.
Observability stacks
One row per Prometheus, Loki, or similar instance Daalu reads from. Click a row for:
- Connection — how Daalu reaches the stack (direct URL or through a federation tunnel).
- Probe history — recent query latencies and errors.
- Routes — which classes of query Daalu sends here, so you know which stack the Assistant is hitting when you run several.
Observability is how the Assistant gets evidence for “why did this alert fire?” questions. Without at least one stack connected, those answers stay short. The provider reference is Chapter 39.
Tab 2 — Network & servers
This tab is the per-tenant surface for managing your network switches and the lifecycle of your bare-metal servers — storing intended config, showing diffs, deploying with automatic rollback, and provisioning servers from bare metal. The find → suggest → approve → apply loop that drives it is the same human-in-the-loop pattern as the rest of Daalu.
It’s a big enough surface to have its own chapter. Chapter 26 is the full walkthrough; this tab is where you land once it’s been provisioned.
Tab 3 — Onboarding
Onboarding is the wizard that stands up the heavier managed components. The main flow here is Network & server management — choosing where the management stack runs (Daalu’s cluster or your own), which components you need, and what size, then provisioning it. When it finishes you get a set of private, per-organization web addresses for each component.
The full wizard, the profiles, and what you need to provide are covered in Chapter 26.
Where the rest of your integrations live
Cloud accounts, your source of truth, notification channels, and webhooks used to share this page; they now live on the Integrations page (Chapter 24), with the per-provider deep dives in Part VI (Chapters 36–40). If you came here looking for “add an AWS account” or “connect Slack,” head to Integrations.