Skip to content

valleyelectronics/cachet-monitor

 
 

Repository files navigation

screenshot

Features

  • Creates & Resolves Incidents
  • Posts monitor lag to cachet graphs
  • HTTP Checks (body/status code)
  • DNS Checks
  • Updates Component to Partial Outage
  • Updates Component to Major Outage if already in Partial Outage (works with distributed monitors)
  • Can be run on multiple servers and geo regions

Example Configuration

Note: configuration can be in json or yaml format. example.config.json, example.config.yaml files.

api:
  # cachet url
  url: https://demo.cachethq.io/api/v1
  # cachet api token
  token: 9yMHsdioQosnyVK4iCVR
  insecure: false
# https://golang.org/src/time/format.go#L57
date_format: 02/01/2006 15:04:05 MST
slack_webhook: https://hooks.slack.com/services/T00000000/B00000000/XXXXXXXXXXXXXXXXXXXXXXXX
monitors:
  # http monitor example
  - name: google
    # test url
    target: https://google.com
    # strict certificate checking for https
    strict: true
    # HTTP method
    method: POST

    # set to update component (either component_id or metric_id are required)
    component_id: 1
    # set to post lag to cachet metric (graph)
    metric_id: 4

    # custom templates (see readme for details)
    # leave empty for defaults
    template:
      investigating:
        subject: "{{ .Monitor.Name }} - {{ .SystemName }}"
        message: "{{ .Monitor.Name }} check **failed** (server time: {{ .now }})\n\n{{ .FailReason }}"
      fixed:
        subject: "I HAVE BEEN FIXED"

    # seconds between checks
    interval: 1
    # seconds for timeout
    timeout: 1
    # If % of downtime is over this threshold, open an incident
    threshold: 80
    # Raw data to transmit
    data: "{\"key\":\"value\"}"
    # custom HTTP headers
    headers:
      Authorization: Basic <hash>
    # expected status code (either status code or body must be supplied)
    expected_status_code: 200
    # regex to match body
    expected_body: "P.*NG"
    # expected body response md5 checksum
    expected_md5sum: "<md5sum>"
    # expected body response length
    expected_length: 12345
  # dns monitor example
  - name: dns
    # fqdn
    target: matej.me.
    # question type (A/AAAA/CNAME/...)
    question: mx
    type: dns
    # set component_id/metric_id
    component_id: 2
    # poll every 1s
    interval: 1
    timeout: 1
    # custom DNS server (defaults to system)
    dns: 8.8.4.4:53
    answers:
      # exact/regex check
      - regex: [1-9] alt[1-9].aspmx.l.google.com.
      - exact: 10 aspmx2.googlemail.com.
      - exact: 1 aspmx.l.google.com.
      - exact: 10 aspmx3.googlemail.com.
  # example tcp
  - name: smtpnine
    target: smtp.nine.ch
    type: tcp
    port: 25
    component_id: 4

Installation

  1. Download binary from release page
  2. Add the binary to an executable path (/usr/bin, etc.)
  3. Create a configuration following provided examples
  4. cachet-monitor -c /etc/cachet-monitor.yaml

pro tip: run in background using nohup cachet-monitor 2>&1 > /var/log/cachet-monitor.log &, or use a tmux/screen session

Usage:
  cachet-monitor (-c PATH | --config PATH) [--log=LOGPATH] [--name=NAME] [--immediate]
  cachet-monitor -h | --help | --version

Arguments:
  PATH     path to config.json
  LOGPATH  path to log output (defaults to STDOUT)
  NAME     name of this logger

Examples:
  cachet-monitor -c /root/cachet-monitor.json
  cachet-monitor -c /root/cachet-monitor.json --log=/var/log/cachet-monitor.log --name="development machine"

Options:
  -c PATH.json --config PATH     Path to configuration file
  -h --help                      Show this screen.
  --version                      Show version
  --immediate                    Tick immediately (by default waits for first defined interval)
  --restarted                    Get open incidents before start monitoring (if monitor died or restarted)

Environment varaibles:
  CACHET_API      override API url from configuration
  CACHET_TOKEN    override API token from configuration
  CACHET_DEV      set to enable dev logging

Init script

If your system is running systemd (like Debian, Ubuntu 16.04, Fedora, RHEL7, or Archlinux) you can use the provided example file: example.cachet-monitor.service.

  1. Simply put it in the right place with cp example.cachet-monitor.service /etc/systemd/system/cachet-monitor.service
  2. Then do a systemctl daemon-reload in your terminal to update Systemd configuration
  3. Finally you can start cachet-monitor on every startup with systemctl enable cachet-monitor.service! 👍

Templates

This package makes use of text/template. Default HTTP template

The following variables are available:

Root objects Description
.SystemName system name
.API api object from configuration
.Monitor monitor object from configuration
.now formatted date string
Monitor variables
.Name
.Target
.Type
.Strict
.MetricID
...

All monitor variables are available from monitor.go

Vision and goals

We made this tool because we felt the need to have our own monitoring software (leveraging on Cachet). The idea is a stateless program which collects data and pushes it to a central cachet instance.

This gives us power to have an army of geographically distributed loggers and reveal issues in both latency & downtime on client websites.

Package usage

When using cachet-monitor as a package in another program, you should follow what cli/main.go does. It is important to call Validate on CachetMonitor and all the monitors inside.

API Documentation

Contributions welcome

We'll happily accept contributions for the following (non exhaustive list).

  • Implement ICMP check
  • Implement TCP check
  • Any bug fixes / code improvements
  • Test cases

Build on MacOS

  1. Read and install with https://ahmadawais.com/install-go-lang-on-macos-with-homebrew/

  2. Test in console with go get -u and go build cli/main.go

  3. Run ./go-executable-build.sh cli/main.go

  4. mv cli/main.go-linux-amd64 cachet-monitor-linux-amd64

Packages

No packages published

Languages

  • Go 95.6%
  • Shell 3.2%
  • Dockerfile 1.1%
  • Makefile 0.1%