Quickstart
Try everything in Docker
Section titled “Try everything in Docker”Prefer not to install a binary on your computer? Start the complete demo in Docker:
curl -fsSL https://kanshi.dev/demo.sh | shThe script downloads the demo into kanshi-demo/, generates private keys, pulls the published demo images, and starts the stack. It does not install the Kanshi Agent on your host.
Open http://localhost:3000 and enter the printed dashboard key. Within the demo you can inspect:
- CPU, memory, disk, and network history on Fleet and agent details.
- The process explorer, enabled explicitly for the demo Agent.
- Healthy, slow, declined, and failed requests on Services.
- Host links from Services and traces into the matching Agent.
- Distributed checkout-to-payments traces and correlated logs.
- A firing memory alert with delivered webhook history.
- Persistent Light and Dark dashboard themes from the Admin menu.
If you prefer to inspect each command first:
git clone https://github.com/kanshi-dev/demo.gitcd demomake upVerify
Section titled “Verify”cd kanshi-demomake verifyThis sends one request through the Go and Node.js sample services and verifies its trace, correlated logs, host and process metrics, component versions, authentication, limits, and graceful shutdown. See Application observability to instrument your own services.
Monitor the actual host
Section titled “Monitor the actual host”The containerized Agent reports its Docker environment. To collect accurate metrics from the computer itself, install the Agent only when you are ready:
curl -fsSL https://kanshi.dev/install.sh | shRun make agent-env in kanshi-demo/ to print the ingest configuration. Copy those values to the monitored machine, replace the address with the demo host, and start kanshi-agent.
To run it as a background service or install on a remote host, see Installation.
Troubleshooting
Section titled “Troubleshooting”401 unauthorized: use the dashboard key for REST and the ingest key for the Agent and Collector.- Degraded health: run
make logs. - Agent or Collector connection failure: confirm
50051is reachable and omit URL schemes fromKANSHI_CORE_ADDR. - Dashboard load failure: clear the stored key and enter
KANSHI_DASHBOARD_KEYagain. - Existing demo volume: restore the original
kanshi-demo/.env, or rundocker volume rm kanshi-demo_kanshi_dbto delete the old database before reinstalling.
Use make down to stop the stack. Run make reset to delete local metrics and regenerate keys on the next start.
The demo commands are maintained in the demo repository.