fujiwara/maprobe repository overview

usage: maprobe [<flags>] <command> [<args> ...]

Flags:
  --help              Show context-sensitive help (also try --help-long and --help-man).
  --log-level="info"  log level

Commands:
  help [<command>...]
    Show help.

  agent [<flags>]
    Run agent

  once [<flags>]
    Run once

  ping [<flags>] <address>
    Run ping probe

  tcp [<flags>] <host> <port>
    Run TCP probe

  http [<flags>] <url>
    Run HTTP probe

$ maprobe agent --help
usage: maprobe agent [<flags>]

Run agent

Flags:
      --help              Show context-sensitive help (also try --help-long and --help-man).
      --log-level="info"  log level
  -c, --config=CONFIG     configuration file path or URL(http|s3)

post_probed_metrics: false   # when false, do not post host metrics to Mackerel. only dump to [info] log.
probes:
  - service: '{{ env "SERVICE" }}'   # expand environment variable
    role: server
    ping:
      address: '{{ .Host.IPAddresses.eth0 }}'

  - service: production
    role: webserver
    http:
      url: 'http://{{ .Host.CustomIdentifier }}/api/healthcheck'
      post: POST
      headers:
        Content-Type: application/json
      body: '{"hello":"world"}'
      expect_pattern: 'ok'

  - service: production
    role: redis
    tcp:
      host: '{{ .Host.IPAddress.eth0 }}'
      port: 6379
      send: "PING\n"
      expect_pattern: "PONG"
      quit: "QUIT\n"
    command:
      command:
        - "mackerel-plugin-redis"
        - "-host={{ .Host.IPAddress.eth0 }}"
        - "-tempfile=/tmp/redis-{{ .Host.ID }}"

ping:
  address: "192.168.1.1"      # Hostname or IP address (required)
  count: 5                    # Iteration count (default 3)
  timeout: "500ms"            # Timeout to ping response (default 1 sec)
  metric_key_prefix:          # default ping

tcp:
  host: "memcached.example.com" # Hostname or IP Address (required)
  port: 11211                   # Port number (required)
  timeout: 10s                  # Seconds of timeout (default 5)
  send: "VERSION\n"             # String to send to the server
  quit: "QUIT\n"                # String to send server to initiate a clean close of the connection"
  expect_pattern: "^VERSION 1"  # Regexp pattern to expect in server response
  tls: false                    # Use TLS for connection
  no_check_certificate: false   # Do not check certificate
  metric_key_prefix:            # default tcp

http:
  url: "http://example.com/"     # URL
  method: "GET"                  # Method of request (default GET)
  headers:                       # Map of request header
    Foo: "bar"
  body: ""                       # Body of request
  expect_pattern: "ok"           # Regexp pattern to expect in server response
  timeout: 10s                   # Seconds of request timeout (default 15)
  no_check_certificate: false    # Do not check certificate
  metric_key_prefix:             # default http

command:
  command: "/path/to/metric-command -option=foo" # execute command
  timeout: "5s"                      # Seconds of command timeout (default 15)
  graph_defs: true                   # Post graph definitions to Mackerel (default false)

command:
  command:
    - "/path/to/metric-command"
    - "-option=foo"
  timeout: "5s"                      # Seconds of command timeout (default 15)
  graph_defs: true                   # Post graph definitions to Mackerel (default false)

Command probe handles command's output as host metric.

When `graph_defs` is true, maprobe runs a command with `MACKEREL_AGENT_PLUGIN_META=1` environment variables and post graph definitions to Mackerel at first time.

If the command does not return a valid graph definitions output, that is ignored.

See also [ホストのカスタムメトリックを投稿する - Mackerel ヘルプ](https://mackerel.io/ja/docs/entry/advanced/custom-metrics#graph-schema).

#### Example of automated cleanup for terminated EC2 instances.

Command probe can run any scripts against for Mackerel hosts.

For example,

```yaml
service: production
role: server
statues:
  - working
  - standby
  - poweroff
command:
  command: 'cleanup.sh {{.Host.ID}} {{index .Host.Meta.Cloud.MetaData "instance-id"}}'

#!/bin/bash
set -u
host_id="$1"
instance_id="$2"
exec 1> /dev/null # dispose stdout
result=$(aws ec2 describe-instance-status --instance-id "${instance_id}" 2>&1)
if [[ $? == 0 ]]; then
  exit
elif [[ $result =~ "InvalidInstanceID.NotFound" ]]; then
   mkr retire --force "${host_id}"
fi

post_aggregated_metrics: false   # when false, do not post service metrics to Mackerel. only dump to [info] log.
aggregates:
  - service: production
    role: app-server
    metrics:
      - name: cpu.user.percentage
        outputs:
          - func: sum
            name: cpu.user.sum_percentage
          - func: avg
            name: cpu.user.avg_percentage
      - name: cpu.idle.percentage
        outputs:
          - func: sum
            name: cpu.idle.sum_percentage
          - func: avg
            name: cpu.idle.avg_percentage