One of the most comprehensive collections of production-ready Grafana dashboards on GitHub.
100+ dashboards that answer operational questions — Linux, Kubernetes, Docker, OpenStack, the monitoring stack, databases and cloud — each generated from a spec, with docs, screenshots, alerts and recording rules.
Most Grafana dashboards online are metric dumps — a wall of gauges that look busy and tell you nothing at 3am. These are different: every dashboard leads with the headline signal (saturation, errors, capacity headroom) and is built to answer a real question — is the cluster healthy? which hypervisor is full? are the queues growing? which disk fills first?
⭐ If this saves you time in an incident, star the repo — it helps other engineers find it.
- Answers questions, not metric dumps. USE for resources, RED for services; the first row tells on-call whether to dig in.
- Generated, so it's consistent. Every dashboard is compiled from a compact
spec — identical units, thresholds, templating and portable
datasource handling (
${DS_PROMETHEUS}/${DS_LOKI}, never a hardcoded UID). - Batteries included. Each dashboard ships a doc page, an annotated screenshot, recommended alerts, and a Production lessons note. Plus a PromQL cookbook, recording rules and datasource provisioning.
- Backend-agnostic. Standard PromQL runs on Prometheus, VictoriaMetrics and Mimir; Loki for logs, Tempo for traces.
- CI-validated. A strict validator enforces schema, units, datasource templating and "no URLs/credentials in dashboard JSON" on every commit.
git clone https://github.com/devopsaitoolkit/grafana-dashboards.git
cd grafana-dashboards
# Import one dashboard via the UI: Dashboards → Import → upload a JSON, pick your datasource.
# Or via the API:
export GRAFANA_URL=https://grafana.example.com GRAFANA_TOKEN=glsa_xxx
scripts/import-dashboard.sh dashboards/linux/cpu.json
# Or provision the whole library as code:
scripts/provision.sh /var/lib/grafana/dashboardsSee docs/provisioning.md and docs/prometheus-setup.md.
| Area | Contents |
|---|---|
📊 dashboards/ |
100+ dashboards by domain (generated JSON) |
🧩 tools/ |
The dashgen generator + the YAML specs (source of truth) |
🚨 alerts/ · recording-rules/ |
Prometheus alert & recording rules |
🔌 datasources/ |
Grafana datasource provisioning (Prometheus/VM/Mimir/Loki/Tempo) |
📚 promql/ |
A PromQL cookbook — 90+ explained queries |
📖 docs/ |
Design, provisioning, variables, alerting, performance + guides |
🖼️ screenshots/ |
Annotated schematic of every dashboard |
🔧 scripts/ |
import, export, backup, restore, diff, provision, screenshot |
Browse the full, always-up-to-date list in docs/catalog.md and the compatibility matrix. Highlights:
| Domain | Examples |
|---|---|
| Linux | CPU, memory, disk I/O, filesystem, network, load, systemd, PSI pressure |
| Kubernetes | cluster overview, nodes, pods, workloads, API server, etcd, scheduler, CoreDNS, ingress, PV/PVC |
| Docker · cAdvisor | containers, engine hosts, images/volumes, networks, compose stacks |
| OpenStack | Nova hypervisor capacity, VM density, Neutron, Cinder, Glance, Keystone, RabbitMQ, MariaDB/Galera, HAProxy, Placement, Heat, OVS |
| Monitoring | Prometheus, Alertmanager, VictoriaMetrics, Mimir, Loki, Tempo |
| Databases | PostgreSQL, MySQL/InnoDB, Redis (keyspace, replication) |
| Web | NGINX, Apache |
| Cloud | VMware vSphere, AWS EC2/RDS, Azure VMs, GCP Compute/GKE |
flowchart LR
S[tools/specs/*.yaml<br/>compact, high-signal] --> G[dashgen]
G --> J[dashboards/*.json]
G --> D[docs + screenshots]
G --> A[alerts/*.rules.yml]
G --> C[catalog + matrix]
You edit a spec, run make build, and the generator produces the dashboard JSON,
its documentation, an annotated screenshot, and Prometheus alert rules — all
consistent by construction. See docs/authoring-specs.md.
pip install pyyaml jsonschema
make build # compile specs → dashboards, docs, screenshots, alerts
make validate # strict checks (schema, units, datasource templating, no URLs)SREs · Platform & DevOps Engineers · Cloud & Infrastructure Engineers · OpenStack operators · Kubernetes administrators · Observability teams.
New dashboards from real operators are what make this library great — adding one is ~20 minutes of editing a spec. See CONTRIBUTING.md, the roadmap, and open a dashboard request. By participating you agree to our Code of Conduct.
Apache-2.0. Use the dashboards, rules and tooling freely.
This repository stands on its own. To go deeper on AI-assisted observability, these free resources from the maintainers help:
- 📚 Advanced observability guides
- 🤖 AI Incident Response Assistant
- 📈 Grafana & Prometheus tutorials
- ✉️ Weekly DevOps newsletter
Built by engineers who stare at these dashboards during real incidents. If it helps you, give it a ⭐.