HTTP server ​

Run chartjs2img serve to expose the same renderer over HTTP. This is the right mode when multiple clients render charts, when renders benefit from a shared in-memory cache, or when you want a single long-running process rather than per-chart subprocess startup.

chartjs2img serve --port 3000

On startup the server prints:

chartjs2img v0.2.2 listening on http://0.0.0.0:3000
  POST /render      - render chart from JSON body
  GET  /render      - render chart from query params
  GET  /cache/:hash - retrieve cached image
  GET  /examples    - examples gallery
  GET  /health      - health check + stats

Press Ctrl-C (or send SIGTERM) to shut down cleanly — the server drains in-flight requests before closing Chromium.

A quick taste of what the POST produces — flip to the HTTP tab to see the exact curl that generates the preview:

PreviewJSONCLIHTTPPOST /render output.

Server flags ​

Other knobs (concurrency, cache, timeouts) are configured via environment variables only — see Environment variables.

POST /render ​

Render a chart from a JSON body.

curl -X POST http://localhost:3000/render \
  -H 'Content-Type: application/json' \
  -d '{
    "chart": {
      "type": "bar",
      "data": { "labels": ["Jan","Feb"], "datasets": [{"data":[1,2]}] }
    },
    "width": 800,
    "height": 600,
    "format": "png"
  }' \
  -o chart.png

Request body ​

Missing / invalid input returns 400 with a JSON error body:

{ "error": "Missing or invalid required field: chart (must be an object)" }

Response headers ​

See Error feedback for how to interpret X-Chart-Messages, and Cache for how the hash is computed.

GET /render ​

Same semantics as POST, but all parameters come from query strings. Useful for embedding in <img> tags.

GET /render?chart={"type":"bar","data":{...}}&width=400&height=300

The chart query parameter must be a URL-encoded JSON string. Beyond a few hundred data points you'll want POST — most clients cap URL length well below chartjs2img's server limit.

GET /cache/:hash ​

Re-fetch a previously rendered image by its cache hash.

# Render and grab the hash
HASH=$(curl -s -D- -X POST http://localhost:3000/render \
  -H 'Content-Type: application/json' \
  -d '{"chart":{"type":"bar","data":{"labels":["A","B"],"datasets":[{"data":[1,2]}]}}}' \
  -o /dev/null | grep -i x-cache-hash | awk '{print $2}' | tr -d '\r')

# Fetch the cached image later
curl -o chart.png "http://localhost:3000/cache/$HASH"

Cached entries expire after CACHE_TTL_SECONDS (default 3600). Missing or expired hashes return 404. Details in Cache.

GET /health ​

Returns server status, renderer stats, cache info, and concurrency counters. Useful for liveness / readiness probes.

{
  "status": "ok",
  "version": "0.2.2",
  "renderer": {
    "browserConnected": true,
    "concurrency": { "max": 8, "active": 2, "pending": 0 },
    "activePages": 2,
    "maxRenderTimeSeconds": 30,
    "pageSafetyNetSeconds": 70,
    "pageTimeoutSeconds": 70
  },
  "cache": {
    "size": 42,
    "maxEntries": 1000,
    "ttlSeconds": 3600
  }
}

pageTimeoutSeconds is a deprecated alias of pageSafetyNetSeconds retained for clients shipped before the field was renamed.

GET /examples ​

Built-in gallery showing every example chart rendered live. The page displays each chart plus its source JSON — a convenient way to verify the service is up and browse available chart types.

Capacity ​

Incoming renders wait on a semaphore with CONCURRENCY slots (default 8). Extra requests queue until a slot frees up. Tune via Environment variables.

Where to next ​

• Cache — hash-based caching mechanism and CDN-friendly URLs.
• Authentication — optional API key setup.
• Docker — container image, docker-compose, reverse proxy.
• Environment variables — every tunable for the server.
• Error feedback — X-Chart-Messages and HTTP error codes.