> ## Documentation Index
> Fetch the complete documentation index at: https://docs.skyhook.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Observability Bundle
> Complete observability stack with metrics, logs, and traces for Kubernetes clusters
## What is the Observability Bundle?
The Observability Bundle is a curated collection of industry-standard tools that provide comprehensive monitoring, logging, and tracing for your Kubernetes clusters. Skyhook manages the installation and configuration of these tools through GitOps workflows, giving you a complete observability solution with minimal setup effort.
The bundle implements the three pillars of observability:
* **Metrics**: Numerical measurements of system performance over time (CPU usage, request rates, error rates)
* **Logs**: Text records of discrete events that occurred in your system
* **Traces**: Records of requests as they flow through distributed services
## Bundle Components
The Observability Bundle includes 9 integrated components that work together to provide complete visibility into your Kubernetes infrastructure and applications.
### Visualization and Dashboards
Central dashboard for visualizing metrics, logs, and traces. Grafana connects to all other components in the bundle to provide unified observability dashboards.
* Multi-datasource support (Prometheus, Loki, Tempo, Mimir)
* Pre-built dashboards for Kubernetes monitoring
* Custom dashboard creation
* Alerting and notifications
### Distributed Tracing
High-volume distributed tracing backend for storing and querying traces from your applications.
* Cost-effective trace storage (object storage backend)
* Integrates with Grafana for trace visualization
* Compatible with OpenTelemetry, Jaeger, and Zipkin
* Search traces by service, duration, tags
Vendor-neutral telemetry collection agent that receives, processes, and exports traces, metrics, and logs.
* Deployment mode: Collector (aggregation)
* Receives telemetry from applications
* Routes data to Tempo, Loki, and other backends
* Protocol support: OTLP, Jaeger, Zipkin
Node-level OpenTelemetry collector running as a DaemonSet (one pod per node).
* Deployment mode: DaemonSet (per-node)
* Collects node and pod-level telemetry
* Lower network overhead for local collection
* Forwards to central collector or backends
### Log Aggregation
Log aggregation system designed for efficiency and ease of use, inspired by Prometheus.
* Cost-effective log storage (indexes metadata, not content)
* LogQL query language (similar to PromQL)
* Integrates seamlessly with Grafana
* Label-based log filtering and search
### Metrics Collection and Storage
Industry-standard metrics collection and time-series database for Kubernetes.
* Automatic service discovery in Kubernetes
* PromQL query language
* Pull-based metrics collection
* Short-term metrics storage (15-30 days typical)
Long-term metrics storage backend compatible with Prometheus.
* Horizontally scalable metrics storage
* Long-term retention (months to years)
* Prometheus remote write compatible
* Multi-tenancy support
Generates metrics about Kubernetes object states (deployments, pods, nodes, etc.).
* Cluster-level resource metrics
* Object state and health information
* Deployment status, pod status, node conditions
* Scraped by Prometheus
Exports hardware and OS-level metrics from Kubernetes nodes.
* CPU, memory, disk, network metrics
* Runs as DaemonSet (one per node)
* System-level performance data
* Node health monitoring
## How Components Work Together
The observability bundle components integrate to provide end-to-end visibility:
```
Application
├─> OpenTelemetry SDK (instrumented code)
│ ├─> Traces ────────────> OTel Collector ────────> Grafana Tempo
│ ├─> Logs ──────────────> OTel Collector ────────> Grafana Loki
│ └─> Metrics ───────────> OTel Collector ────────> Prometheus/Mimir
│
Kubernetes Cluster
├─> Prometheus ──────────────> Scrapes metrics from:
│ ├─> Kube State Metrics (cluster objects)
│ ├─> Node Exporter (node resources)
│ └─> Application /metrics endpoints
│ └─> Remote Write ────────> Grafana Mimir (long-term storage)
│
└─> Grafana ─────────────────> Visualizes data from:
├─> Prometheus (real-time metrics)
├─> Mimir (historical metrics)
├─> Loki (logs)
└─> Tempo (traces)
```
## Installation
The Observability Bundle is installed and managed through Skyhook's GitOps integration using ArgoCD.
### Prerequisites
* ArgoCD installed and configured (see [GitOps Setup](/infrastructure/gitops/overview))
* At least one Kubernetes cluster connected to Skyhook
* Management cluster identified (where observability backends will run)
### Installing the Bundle
1. Navigate to **Addons** in the Skyhook dashboard
2. Click on **Observability Bundle**
3. Click **Configure** to set up the bundle
The configuration wizard guides you through:
* Selecting management and workload clusters
* Configuring storage settings
* Setting up ingress for Grafana and OpenTelemetry
* Managing secrets (Grafana credentials, authentication tokens)
### Architecture Patterns
**Management Cluster**: Runs the observability backends
* Grafana (visualization)
* Tempo (trace storage)
* Loki (log storage)
* Mimir (metrics storage)
* Prometheus (metrics collection)
**Workload Clusters**: Run data collection agents
* OpenTelemetry Collector (send traces/logs to management cluster)
* Prometheus (scrape local metrics, remote write to management cluster)
* Kube State Metrics (expose cluster metrics)
* Node Exporter (expose node metrics)
This centralized architecture allows you to monitor all your clusters from a single Grafana instance.
## Component Documentation
For detailed setup and usage of individual components:
* [Metrics](/observability/metrics) - Prometheus, Mimir, Kube State Metrics, Node Exporter
* [Logs](/observability/logs) - Grafana Loki log aggregation
* [Tracing](/observability/tracing) - Grafana Tempo and OpenTelemetry distributed tracing
## Benefits
Single pane of glass for metrics, logs, and traces across all your clusters
Infrastructure as code with version control, audit trails, and easy rollbacks
Best-of-breed open source tools used by thousands of organizations
Efficient storage and retention strategies minimize infrastructure costs
Monitor all your Kubernetes clusters from a centralized observability stack
OpenTelemetry-based collection works with any instrumented application
## Next Steps
Set up GitOps and install the Observability Bundle
Set up Prometheus and Mimir for metrics collection
Configure Loki for centralized log management
Implement tracing with Tempo and OpenTelemetry