2. Your first edge instance and your first app
30-60 minutes. By the end you'll have a Kubernetes cluster registered in DT Edge Platform and one application installed on it from the marketplace.
What you need before starting
- A Kubernetes cluster to manage (yours or one you have a kubeconfig for)
- The kubeconfig file for that cluster
- An organization selected in the sidebar
- The
edge-instances:createpermission in that org (default member role usually has this — if not, ask your admin)
Step 1 — Add your edge instance
In the sidebar, click Edge Instances → click Add Edge at the top right.
Two paths show up:
- Upload kubeconfig — paste or drag in a kubeconfig file. Use this when the cluster already exists.
- Provision new cluster — only available when your install has the upstream Provisioner integration configured. Skip this for the tutorial.
Pick Upload kubeconfig. Fill in:
- Name — what you'll call this cluster (e.g.
staging-istanbul). Pick something you'll recognise in lists. - Description (optional) — a sentence about its purpose.
- Categories (optional) — tags that help you filter the list later.
- Kubeconfig — paste the YAML or upload the file.
Submit. DT Edge Platform encrypts the kubeconfig with its master key, attempts a probe against the cluster, and the new edge appears in the list. Status badges to expect:
- 🟡
in_progressfor a few seconds while DT Edge Platform probes - 🟢
activeonce the probe succeeds - 🔴
failedif the probe couldn't reach the cluster — open the edit drawer and check the error in the health section
Step 2 — Open the marketplace
Sidebar → Marketplace.
You'll see a grid of available packages. Each card shows:
- The package name + icon
- Categories (you can filter by these in the top bar)
- A Versions count
If your install hasn't published any packages yet, the grid will be empty. In that case, ask your admin to publish at least one, or skip to Tutorial 3 and come back when you have a package to install.
Step 3 — Install an app
Click any package card → click Install.
The install drawer opens with three sections:
- Target — pick the edge instance you just added (Step 1). The dropdown also lets you pick a namespace and a release name. Defaults are sensible if you don't know what to put.
- Version — most packages have one or more versions. Pick the latest unless you have a reason not to.
- Values — the chart's configuration. You'll see one of two
editors:
- Form view (when the package author shipped a typed form schema) — friendly fields, validation, dropdowns
- YAML view — the raw values.yaml the chart expects
Toggle between the two with the button at the top of this section. Form view is the default when available; switch to YAML if you want to see what the form is generating, or if you need a field the form doesn't expose.
Click Install. The drawer closes; you'll see a toast notification "Operation queued". A new row appears in the Applications sidebar entry.
Step 4 — Watch the install land
Sidebar → Applications. Find the row whose release name you just used.
The status column moves through:
- ⏳
pending— queued, worker hasn't picked it up - 🔄
running/installing— helm install is executing - ✅
deployed— done; the chart's resources are on the cluster
Click the row → Detail to see live progress. The detail page streams logs over WebSocket; you don't need to refresh.
If the status flips to failed, click into Detail → look at the
Last error card. Most failures are one of:
- Chart pull failed (registry unreachable / wrong creds)
- A required value missing (form validation should catch this, but YAML edits can slip through)
- The cluster's API server unreachable (back to Edge Instances → health)
For deeper diagnosis, see the Helm install failed runbook in the backend docs.
Step 5 — Check it on the cluster (optional)
If you have kubectl access to the cluster, sanity-check what
landed:
kubectl --kubeconfig=<your-config> -n <namespace> get all -l app.kubernetes.io/managed-by=Helm
You should see the resources the chart created. DT Edge Platform marks the
release with a dtedge.io/managed=true label so it'll only show
up in your Applications list (not, say, kube-system pods that
were on the cluster before you registered it).
What's next
You have a managed cluster + an app running on it. Useful follow-ups:
- Tutorial 3 — Set up your first alarm — get notified when something breaks
- How-to: upgrade or rollback an application
- How-to: manage edge instances — node listing, health, removing edges