About collectd metrics
Collectd is a daemon that collects system performance statistics periodically and stores them in a variety of ways. For GitHub Enterprise Server, collectd gathers metrics from various system components and services, providing insight into system health and performance.
Key components
The collectd monitoring stack includes the following components:
- Collectd: The main daemon that collects system performance statistics
- Graphite: Serves as the datasource for dashboard visualizations
Metrics collection
Collectd gathers metrics from various sources including:
- System metrics: CPU usage, memory utilization, disk I/O, network statistics
- Application metrics: HAProxy statistics, Resque queue metrics, database performance
- Custom metrics: Service-specific metrics through custom plugins and scripts
Architecture
Single appliance
In a single appliance deployment, collectd runs locally and stores metrics in RRD (Round Robin Database) files. The 管理控制台 reads these files to display monitoring dashboards.
Cluster environment
In cluster environments, collectd operates in a distributed manner:
- Metrics servers: Designated nodes that collect and store metrics from all cluster nodes
- Metrics clients: All other nodes that forward their metrics to the metrics servers
- Redundancy: Metrics are duplicated across multiple metrics servers for failover support
Configuring collectd metrics
Collectd metrics are enabled by default on GitHub Enterprise Server instances.
External monitoring with collectd
You can set up external monitoring systems to collect and analyze collectd metrics from your GitHub Enterprise Server instance. This enables integration with existing monitoring infrastructure and provides additional visualization and alerting capabilities.
For more information about setting up external monitoring, see Setting up external monitoring with collectd.
Collectd metrics reference
GitHub Enterprise Server collects various metrics through collectd, covering system resources, application performance, and service health. Understanding these metrics is essential for effective monitoring and troubleshooting.
For a comprehensive list of available metrics, see collectd metrics for GitHub Enterprise Server.
Next steps
- To set up external monitoring, see Setting up external monitoring with collectd.
- To explore available metrics, see collectd metrics for GitHub Enterprise Server.