"Agent unreachable" — what to check
Empty lists and amber warning banners almost always mean the console cannot reach the GraphVM agent.
Pages that read live data show an amber banner instead of content when their /api/* call fails. The console renders fine; it just has nothing to show. Work through the checks below in order.
- Confirm the agent is running and serving its health endpoint. Workspace settings shows agent health (status, engine, version) at the top when the call succeeds.
- Check the URL the console proxies to. It defaults to http://127.0.0.1:9090 and is overridden with GRAPHVM_AGENT_INTERNAL_API.
- Open the agent lab at /dev/agent to run health checks and probes directly.
- If health passes but one page still fails, the problem is that specific endpoint — check the agent logs for the route named in the banner.
export GRAPHVM_AGENT_INTERNAL_API=http://127.0.0.1:9090
yarn devRelated
- What GraphVM isGetting startedGraphVM runs workflows you draw as graphs; this console is the editor and control room in front of the agent that executes them.
- Workspace settings and defaultsWorkspace & adminTenant-wide defaults — theme, default graph scope, experimental nodes, editor grid snap — plus agent health.
More in Getting started · Search all help