LiteMon

A very minimal and lightweight metric collector for Linux systems.

LiteMon allows lightweight monitoring of Linux nodes (VM, bare-metal, etc.). LiteMon is carefully crafted to use as little resources as possible.

LiteMon is using less than 5 MiB of RAM (99th percentile), the binary size is less than 10 MB, and uses very little I/O and CPU. Binaries are built with the musl libc and have no external dependencies.

LiteMon is written in Rust, and uses the Smol async runtime.

Installation

Latest version:

0.1.2

Ubuntu/Debian

# Install litemon
sudo dpkg -i $(curl -w "%{filename_effective}" -SsLO "https://github.com/RagnarLab/litemon/releases/download/v0.1.2/litemon-0.1.2-1.$(uname -m).deb")

# Configure litemon
sudo mv /etc/litemon/config.kdl.example /etc/litemon/config.kdl
sudo $EDITOR /etc/litemon/config.kdl

# Restart litemon
sudo systemctl restart litemon

RHEL/Fedora/CentOS Stream

# Install litemon
sudo rpm -iv "https://github.com/RagnarLab/litemon/releases/download/v0.1.2/litemon-0.1.2-1.$(uname -m).rpm"

# Configure litemon
sudo mv /etc/litemon/config.kdl.example /etc/litemon/config.kdl
sudo $EDITOR /etc/litemon/config.kdl

# Restart litemon
sudo systemctl restart litemon

Configuration

By default, litemon reads the configuration from /etc/litemon/config.kdl. The configuration is written in KDL.

metrics {
  cpu_seconds enabled=#true period_ms=200

  loadavg enabled=#true

  memory_used enabled=#true

  systemd_unit_state enabled=#true {
    // List all the units to monitor.
    units "docker.service"
  }

  network_throughput enabled=#true {
    // List all interfaces to monitor
    interfaces \
        "eth0" \
        "lo"
  }

  disk_usage enabled=#true {
    mountpoints "/"
  }
}

CLI

Usage: litemon [OPTIONS] [PATH-TO-CONFIG]

Options:
-n, --listen          IP address to listen. Default: 127.0.0.1
-P, --port            Port to listen. Default: 9774
-V, --version         Print version info and exit
-h, --help            Print help and exit

Using with alloy

You can easily use litemon with Grafana Alloy.

// config.alloy

// https://grafana.com/docs/alloy/latest/reference/components/discovery/discovery.relabel/
discovery.relabel "integrations_litemon_exporter" {
  targets = [
    // Optionally, update to your listen address and port.
    {"__address__" = "localhost:9774"},
  ]

  rule {
    target_label = "instance"
    replacement  = constants.hostname
  }

  rule {
    target_label = "job"
    replacement  = "integrations/litemon"
  }
}

// https://grafana.com/docs/alloy/latest/reference/components/prometheus/prometheus.scrape/
prometheus.scrape "litemon_exporter" {
  job_name        = "integrations/litemon"
  targets         = discovery.relabel.integrations_litemon_exporter.output
  scheme          = "http"
  scrape_interval = "60s"
  metrics_path    = "/metrics"

  // Update endpoint to your prometheus ingest.
  forward_to = [prometheus.remote_write.EXAMPLE.receiver]
}

Metrics

Metric Name	Metric Type	Description	Cardinality
litemon_load_avg_1m	Gauge	Load average over 1 minute.	1 per host
litemon_load_avg_5m	Gauge	Load average over 5 minutes.	1 per host
litemon_load_avg_15m	Gauge	Load average over 15 minutes.	1 per host
litemon_cpu_usage_overall	Gauge	Overall CPU usage percentage (0.0-1.0).	1 per host
litemon_cpu_usage_per_core	Gauge	Per-core CPU usage percentage (0.0-1.0).	1 per cpu core
litemon_mem_used_percentage	Gauge	Memory used (0.0-1.0) in percent.	1 per host
litemon_systemd_unit_state	Gauge	Systemd unit state (1 for current state, 0 otherwise).	1 per service, 1 per state, 8 states
litemon_net_bytes_received	Counter	Network bytes received.	1 per host, 1 per network interface
litemon_net_errors_received	Counter	Network errors received.	1 per host, 1 per network interface
litemon_net_bytes_sent	Counter	Network bytes sent.	1 per host, 1 per network interface
litemon_net_errors_sent	Counter	Network errors sent.	1 per host, 1 per network interface
litemon_fs_usage_ratio	Gauge	Filesystem usage ratio (0.0-1.0).	1 per host, 1 per mount point

Support

Development of LiteMon is sponsored by RagnarLab. RagnarLab is a Rust consultancy based in Stuttgart, Germany. We provide Rust development from prototype to product, helping you write safer software. Interested in Rust? Get in touch with us.

License

^{Licensed under either of Apache License, Version 2.0 or MIT license at your option.}