sherlock

Sherlock is a Prometheus exporter that collects metrics from Redfish-enabled BMCs, focusing on system, temperature, voltage, power, and fan metrics.

Features

• Collects comprehensive metrics from Redfish-enabled servers
• Supports monitoring multiple servers with a single exporter instance
• Parallel collection of metrics for improved performance
• Detailed health status for most server components

Building

Local Build

go build -o sherlock ./cmd/sherlock

Docker Build

docker build -t sherlock .

Running

Local Run

export REDFISH_USERNAME="admin"
export REDFISH_PASSWORD="password"
./sherlock

Docker Compose

# Set your BMC credentials in environment
export REDFISH_USERNAME="admin"
export REDFISH_PASSWORD="password"

# Run with docker-compose
docker-compose up -d

Configuration

The following environment variables are available:

• REDFISH_USERNAME: BMC username (default: "admin")
• REDFISH_PASSWORD: BMC password (default: "password")
• REDFISH_INSECURE: Allow insecure HTTPS connections (default: true)
• LISTEN_ADDRESS: Address to listen on (default: ":9290")
• METRICS_PATH: Path to expose metrics on (default: "/metrics")

Multi-Server Monitoring

Sherlock requires a target parameter to specify which server to monitor. The target parameter should be just the hostname of the Redfish endpoint (HTTPS is used automatically):

http://sherlock:9290/metrics?target=bmc1.example.com
http://sherlock:9290/metrics?target=bmc2.example.com

If no target is specified, the metrics endpoint will return an error.

In your Prometheus configuration, you can use this feature with relabeling:

scrape_configs:
  - job_name: "redfish"
    scrape_interval: "60s"
    metrics_path: "/metrics"
    params:
      target: ["bmc1.example.com", "bmc2.example.com"]
    static_configs:
      - targets: ["sherlock:9290"]
    relabel_configs:
      - source_labels: [__param_target]
        target_label: "instance"

This allows you to use a single Sherlock instance to monitor multiple servers while keeping the same credentials.

Metrics

The exporter provides the following metrics:

System Metrics

• ipmi_system_power_state: System power state (1 = On, 0 = Off)
• ipmi_cpu_health: CPU health status with model and core count as labels
• ipmi_memory_health: Overall memory subsystem health status with total memory size

Temperature Metrics

• ipmi_temperature_celsius: Temperature readings in Celsius with type labels
• ipmi_temperature_health: Health status of temperature sensors

Voltage Metrics

• ipmi_voltage_volts: Voltage readings in Volts
• ipmi_voltage_health: Health status of voltage sensors

Power Supply Metrics

• ipmi_psu_health: Power supply health status
• ipmi_psu_input_power_watts: Power supply AC input power in Watts
• ipmi_psu_output_power_watts: Power supply DC output power in Watts

Fan Metrics

• ipmi_fan_health: Fan health status
• ipmi_fan_state: Fan state (1 = On, 0 = Off)
• ipmi_fan_speed_rpm: Fan speed in RPM