Linux system metrics collector for statsd written in C++.

Metrics gathered by the application:

• memory (real + virtual) total/usage
• disk space total/usage
• processor usage
• processors count
• network in/out
• disk in/out
• uptime

In statsd they are prefixed with himond.hostname string (or himond.general_hostname if the hostname couldn't be retrieved), e.g.:

himond.production.system.uptime:124|c
himond.production.cpu.usage:77|g

System uptime in seconds.

• system.uptime count

Memory statistics in megabytes.

• memory.total gauge
• memory.used gauge
• memory.free gauge
• memory.virtual.total gauge
• memory.virtual.used gauge
• memory.virtual.free gauge

Disk usage statistics in megabytes, transfer rates in kilobytes per second.

• disk.total gauge
• disk.used gauge
• disk.free gauge
• disk.rate.read count
• disk.rate.write count

Net IO interface transfer rates in kilobytes per second.

• net.rate.recv count
• net.rate.send count

Usage in percent.

• cpu.count count
• cpu.usage count

Himond gathers statistics from the local host and sends them to the selected statsd server. It's necessary to provide the server details where the statistics will be sent (host and port).

You can also customize ethernet interface which will be monitored (default eth0) and the statistics refresh rate in seconds.

Usage:   ./himond <statsd_host> <statsd_port> [<eth_interface>] [<refresh_rate_seconds>]
  where: eth_interface        - default: eth0
         refresh_rate_seconds - default: 1

Example: ./himond 127.0.0.1 8125

Run make from within the src directory. Run make clean to delete all files generated after a build.

A Vagrantfile file is attached that allows to run a test linux environment. It also contains statsd server as a docker container so you can test himond with statsd server just by running himond 127.0.0.1 8125.

Run test_client application to see the metrics on the screen. Run test_client -h for usage.

If Vagrant can't mount folders, try running vagrant plugin install vagrant-vbguest as the problem may be caused by updated linux kernel in the guest system.

• show processor usage separately for each core

MIT