n8n data source for Grafana

Community plugin. Not affiliated with, endorsed by or supported by n8n GmbH. "n8n" is a trademark of n8n GmbH.

Workflow and execution observability for n8n Cloud and self-hosted n8n, based on the n8n public REST API. Configure a URL and an API key; no Prometheus scraping and no database access are needed.

• Execution tables with links into the n8n editor
• Execution statistics as time series: counts by status, success and error rate, duration percentiles, per workflow or in total
• Long-term aggregates from n8n Insights, independent of execution pruning
• Workflow inventory and the n8n security audit
• Template variables for workflows, projects and tags
• Alerting, for example on the error rate of a workflow
• Bundled dashboard n8n Workflow Executions (API)

When to use this plugin

Setup	Recommended
n8n Cloud	This plugin. n8n Cloud offers neither the /metrics endpoint nor database access, so the public API is the only source.
Self-hosted n8n, no extra infrastructure wanted	This plugin: URL and API key are enough.
Self-hosted n8n with Prometheus / Postgres access	Combine with the community dashboards n8n System Health Overview (Prometheus) and n8n Workflow & Execution Analytics (PostgreSQL).

Requirements

• Grafana 13.2 or later
• n8n with the public API enabled
  • n8n Cloud: Starter, Pro or Enterprise plan. The API is not available in the free trial.
  • Self-hosted: all editions. The API is on by default and can be disabled with N8N_PUBLIC_API_DISABLED=true.
• Grafana must reach n8n over the network. Grafana Cloud cannot reach n8n on a private network directly; use a public URL or Private Data Source Connect.

Create an API key

In n8n open Settings → n8n API → Create an API key. Use a dedicated n8n user for Grafana and give the key only read scopes:

workflow:list, workflow:read, execution:list, execution:read, tag:list, project:list, insights:read, securityAudit:generate

Recent n8n versions enforce these scopes on every edition. The plugin itself only sends read requests (enforced by an allowlist in the backend), but a key with write scopes would allow more than the plugin needs.

Configuration

Field	Description
URL	Base URL of the public API including /api/v1. n8n Cloud: https://<instance>.app.n8n.cloud/api/v1, self-hosted: http://<host>:5678/api/v1 (add N8N_PATH if n8n runs under a sub path).
API key	The key created above. Stored encrypted, only used by the Grafana backend.
Skip TLS verify	Do not verify the server certificate. For testing only.
With CA cert / CA cert	Verify against a custom CA, e.g. an internal PKI.
Max pages	Upper limit of API pages (250 executions each) per query. Default 20, range 1–100.
Cache TTL	Seconds API responses are cached in the backend. Default 60, 0 disables caching.
Private data source connect	Grafana Cloud: reach a self-hosted n8n on a private network through a PDC agent. Shown under Additional settings when PDC is enabled for the stack.

Save & test checks the connection and reports whether the n8n version supports the server-side time filter. Typical messages:

• 401: wrong or expired key. n8n answers the same way when the public API is disabled.
• 403: free trial of n8n Cloud, or the key lacks the workflow:list scope.
• 404 / web page instead of API: the URL does not end with /api/v1.
• TLS error: enable With CA cert for internal certificate authorities.
• Cannot reach n8n: the host is not reachable from the Grafana server (or the PDC agent).

The messages do not include the underlying error, so they do not reveal host names, addresses or ports. The Grafana server log has the details (Health check failed).

The URL must not contain a user name or password (https://user:password@host/...): it is stored unencrypted and used for the links in query results. Authenticate with the API key.

Provisioning example:

apiVersion: 1
datasources:
  - name: n8n
    type: nluecke-n8n-datasource
    access: proxy
    url: https://example.app.n8n.cloud/api/v1
    jsonData:
      maxPages: 20
      cacheTtlSeconds: 60
    secureJsonData:
      apiKey: ${N8N_API_KEY}

Query types

Query type	Result	Options
Execution stats	Time series per panel interval or one value for the whole range	Metric (count, success rate, error rate, duration percentiles), group by (status, workflow), filters
Executions	Table: execution, workflow, status, mode, start, stop, duration, retry of, link	Workflows, statuses, project
Workflows	Table: name, active, archived, tags, project, trigger count, "skips successful executions"	Workflows, project
Insights	Summary (with change vs. previous period) or time series per hour/day/week	Mode, bucket, project
Audit	n8n security audit report	–

Success rate = success / (success + error + crashed). Running, waiting and canceled executions are not counted.

Limitations

• Execution retention limits the time range. Statistics and tables are computed from the execution list, which n8n prunes: self-hosted by default after 14 days or 10,000 executions (EXECUTIONS_DATA_MAX_AGE, EXECUTIONS_DATA_PRUNE_MAX_COUNT), on n8n Cloud depending on the plan (e.g. Starter 7 days / 2,500, Pro 30 days / 25,000). Panels show a notice when the time range starts before the oldest stored execution. Use the Insights query for longer ranges.
• Workflows that do not save successful executions (workflow setting, or EXECUTIONS_DATA_SAVE_ON_SUCCESS=none for the whole instance) make counts and success rates too low. The plugin warns for affected workflows; the instance-wide setting cannot be detected through the API.
• Data volume: every query loads the executions of its time range, 250 per request, up to Max pages (default 20 = 5,000 executions). Selected statuses and workflows are requested separately and filtered by n8n, so filtered panels (e.g. failed executions) stay complete on busy instances. If the limit is reached, the panel shows a warning icon with "Incomplete data": how many executions were loaded, which part of the time range they cover, and how to fix it. Clicking the icon opens the data source settings, where you raise Max pages (up to 100). Alternatively narrow the time range or filters, or use the Insights query for totals.
• Older n8n versions have no server-side time filter; the plugin then pages from the newest execution backwards until the start of the time range, which needs more requests for ranges in the past.
• Insights cover all workflows of the instance (or a project), are updated hourly, and their allowed ranges depend on the licence: Community Edition allows ranges of at least 24 hours within the last 7 days; hourly data and longer history need Business or Enterprise (n8n Cloud: by plan).
• Caching: results can be up to Cache TTL seconds old.
• Projects are only available with an n8n licence that includes them; otherwise the project filter is hidden.