Skip to content

install.md

Latest commit

 

History

History
105 lines (79 loc) · 5.12 KB

install.md

File metadata and controls

105 lines (79 loc) · 5.12 KB
Raw

Install

Vantage Studio ships as a single Docker image. The fastest way to see it running is the demo docker compose stack that brings up Studio + OAP + BanyanDB together; for production deployments see docker.md.

Demo: docker compose stack

git clone https://github.com/wu-sheng/vantage-studio
cd vantage-studio
make compose-up

This builds the Studio image locally and runs:

  • studio on :8080
  • oap (apache/skywalking-oap-server:admin-server) on the cluster network — admin port 17128 and status port 12800 are not exposed to the host
  • banyandb storage on its default port

Visit http://localhost:8080 and log in:

  • username: admin
  • password: vantage-changeme

Replace the default password before exposing this instance. The first-run config seed at /data/studio.yaml is hot-reloaded — edit it in place and the BFF picks up the change within ~1 s. See auth.md for the vsadmin:hash workflow.

docker compose down --volumes (or make compose-down) wipes the state.

OAP image expectation

Studio binds to the admin-server module introduced by SWIP-13 — a shared HTTP server (port 17128) that hosts the runtime-rule plugin's REST surface alongside OAL listing and the DSL live debugger. The compose stack expects an OAP image tagged locally as apache/skywalking-oap-server:admin-server built from the SWIP-13 branch. Build it from a checkout of apache/skywalking (./mvnw … && docker build …) or pull from your own CI registry.

The minimum SkyWalking version is 10.5.0 — see compatibility.md.

OAP enablement — four required selectors

Studio's BFF only talks to admin-server-bound OAP modules. All four selectors default to empty (disabled) in stock OAP; Studio requires all four. Set them on the OAP container:

SW_ADMIN_SERVER=default          # shared HTTP server on :17128 (host module for the others)
SW_RECEIVER_RUNTIME_RULE=default # /runtime/rule/* + /runtime/oal/*
SW_DSL_DEBUGGING=default         # /dsl-debugging/* live debugger
SW_INSPECT=default               # /inspect/* — catalog + entity browser (SWIP-14)

What breaks if you omit each one:

Selector What stops working in Studio
SW_ADMIN_SERVER Everything. Without the host module, the other three fail at boot with ModuleNotFoundException: admin-server.
SW_RECEIVER_RUNTIME_RULE DSL Management pages (Catalog, OAL catalog), the Editor's rule fetch/save path, the Cluster status rule-convergence matrix, the Live debugger's rule picker, and the Inspect drawer's source attribution (every metric falls back to the unknown bucket).
SW_DSL_DEBUGGING The Live debugger across all three DSLs (start / poll / stop), and the DSL-debugging health pane in Cluster status.
SW_INSPECT The Inspect page — every /api/inspect/* call returns 404 inspect_not_enabled and the page renders an actionable banner instead of the board.

The "Backend unreachable" / oap_unreachable banners you see in Studio when a selector is missing are honest reports — Studio is running fine, the upstream just isn't exposing the path. Set the selector + restart OAP and the page recovers on the next poll.

The admin-server has no authentication in this release. Reach it only over the cluster's private network — never expose port 17128 to the public internet. Studio is the only authenticated client.

Production

For Docker / Kubernetes deployment patterns, image override knobs, secret layering, and reverse-proxy notes, see docker.md. For the YAML schema see configure.md. For users + RBAC + audit see auth.md.

Helm chart

Deferred to a follow-up release. Until then, the Docker image works in a Deployment + Service + PersistentVolumeClaim of your own — docker.md has a working manifest.