Kapacitor alerts overview

Warning! This page documents an earlier version of Kapacitor, which is no longer actively developed. Kapacitor v1.5 is the most recent stable version of Kapacitor.

Kapacitor makes it possible to handle alert messages in two different ways.

  • The messages can be pushed directly to an event handler exposed through the Alert node.
  • The messages can be published to a topic namespace to which one or more alert handlers can subscribe.

No matter which approach is used, the handlers need to be enabled and configured in the configuration file. If the handler requires sensitive information such as tokens and passwords, it can also be configured using the Kapacitor HTTP API.

Push to handler

Pushing messages to a handler is the basic approach presented in the Getting started with Kapacitor guide. This involves simply calling the relevant chaining method made available through the alert node. Messages can be pushed to log() files, the email() service, the httpOut() cache and many third party services.

Publish and subscribe

An alert topic is simply a namespace where alerts are grouped. When an alert event fires it can be published to a topic. Multiple handlers can subscribe (can be bound) to that topic and all handlers process each alert event for the topic. Handlers get bound to topics through the kapacitor command line client and handler binding files. Handler binding files can be written in yaml or json. They contain four key fields and one optional one.

  • topic: declares the topic to which the handler will subscribe.
  • id: declares the identity of the binding.
  • kind: declares the type of event handler to be used. Note that this needs to be enabled in the kapacitord configuration.
  • match: (optional) declares a match expression used to filter which alert events will be processed. See the section Match Expressions below.
  • options: options specific to the handler in question. These are listed below in the section List of handlers

Example 1: A handler binding file for the slack handler and cpu topic

topic: cpu
id: slack
kind: slack
options:
  channel: '#kapacitor'

Example 1 could be saved into a file named slack_cpu_handler.yaml.

This can then be generated into a Kapacitor topic handler through the command line client.

$ kapacitor define-topic-handler slack_cpu_handler.yaml

Handler bindings can also be created over the HTTP API. See the Create a Handler section of the HTTP API document.

For a walk through on defining and using alert topics see the Using Alert Topics walk-through.

Handlers

A handler takes action on incoming alert events for a specific topic. Each handler operates on exactly one topic.

List of Handlers

The following is a list of available alert handlers and their options.

Aggregate

Aggreate multiple events into a single event.

Options:

Name Type Description
interval duration string How often to aggregate events.
topic string A topic into which to publish the aggregate events.
message string A template string where {{.Interval}} and {{.Count}} are available for constructing a meaning full message.

Example:

kind: aggregate
options:
    interval: 5m
    topic: agg_5m

Send aggregate events of the past 5m to the agg_5m topic. Further handling of the aggregated events can be configured on the agg_5m topic.

Alerta

Send alert events to an Alerta instance.

Options:

Name Type Description
token string Alerta authentication token. If empty uses the token from the configuration.
token-prefix string Alerta authentication token prefix. If empty uses Bearer.
resource string Alerta resource. Can be a template and has access to the same data as the AlertNode.Details property. Default: {{ .Name }}
event string Alerta event. Can be a template and has access to the same data as the idInfo property. Default: {{ .ID }}.
environment string Alerta environment. Can be a template and has access to the same data as the AlertNode.Details property. Defaut is set from the configuration.
group string Alerta group. Can be a template and has access to the same data as the AlertNode.Details property. Default: {{ .Group }}.
value string Alerta value. Can be a template and has access to the same data as the AlertNode.Details property. Default is an empty string.
origin string Alerta origin. If empty uses the origin from the configuration.
service list of string List of effected Services.

Example:

kind: alerta
options:
    resource: system

Exec

Execute an external program, the alert data is passed over STDIN to the process.

Options:

Name Type Description
prog string Path to program to execute.
args list of string List of arguments to the program.

Example:

kind: exec
options:
    prog: /path/to/executable

Hipchat

Send alert events to a Hipchat room.

Options:

Name Type Description
room string HipChat room in which to post messages. If empty uses the channel from the configuration.
token string HipChat authentication token. If empty uses the token from the configuration.

Example:

kind: hipchat
options:
    room: '#alerts'

Log

Log alert events to a file.

Options:

Name Type Description
path string Path to the log file.
mode int File mode to use when creating the file.

Example:

kind: log
options:
    path: '/tmp/alerts.log'

Opsgenie

Send alert events to OpsGenie.

Options:

Name Type Description
teams-list list of string List of teams.
recipients-list List of recipients.

Example:

kind: opsgenie
options:
    teams:
        - rocket

Pagerduty

Send alert events to PagerDuty.

Options:

Name Type Description
service-key string The service key.

Example:

kind: pageduty

Pushover

Send alert events to Pushover.

Options:

Name Type Description
device string Specific list of user’devices rather than all of a user’s devices (multiple device names may be separated by a comma)
title string The message title, otherwise the apps name is used.
url string A supplementary URL to show with the message.
url-title string A title for a supplementary URL, otherwise just the URL is shown.
sound string The name of one of the sounds supported by the device clients to override the user’s default sound choice.

Example:

kind: pushover
options:
    title: Alert from Kapacitor

Post

Post JSON encoded alert data to an HTTP endpoint.

Options:

Name Type Description
url string The URL to which the alert data will be posted.
endpoint string Name of a configured httppost endpoint, cannot be specified in conjunciton with URL.
headers map of string to string Set of extra header values to set on the POST request.

Example:

kind: post
options:
    url: http://example.com

Publish

Publish events to another topic.

Options:

Name Type Description
topics list of string List of topic names to publish events.

Example:

kind: publish
options:
    topics:
        - system
        - ops_team

Sensu

Send alert events to Sensu.

Options:

Name Type Description
source string Sensu source for which to post messages. If empty uses the source from the configuration.
handlers list of string Sensu handler list. If empty uses the handler list from the configuration.

Example:

kind: sensu

Slack

Send alert events to Slack.

Options:

Name Type Description
channel string Slack channel in which to post messages. If empty uses the channel from the configuration.
username string Username of the Slack bot. If empty uses the username from the configuration.
icon-emoji string IconEmoji is an emoji name surrounded in ‘:’ characters. The emoji image will replace the normal user icon for the slack bot.

Example:

kind: slack
options:
    channel: '#alerts'

SMTP

Send alert events via email.

Options:

Name Type Description
to list of string List of email addresses.

Example:

kind: smtp
options:
    to:
        - oncall1@example.com
        - oncall2@example.com

Snmptrap

Trigger SNMP traps for alert events.

Options:

Name Type Description
trap-oid string OID of the trap.
data-list object Each data object has oid, type, and value fields. Each field is a string.

Example:

kind: snmptrap
options:
    trap-oid: '1.1.1.1'
    data-list:
        oid: '1.3.6.1.2.1.1.7'
        type: i
        value: '{{ index .Field "value" }}'

Talk

Send alert events to a Talk instance. No options are available. See the Talk service configuration.

Example:

kind: talk

TCP

Send JSON encoded alert data to a TCP endpoint.

Options:

Name Type Description
address string Address of TCP endpoint.

Example:

kind: tcp
options:
    address: 127.0.0.1:7777

Telegram

Send alert events to a Telegram instance.

Options:

Name Type Description
chat-id string Telegram user/group ID to post messages to. If empty uses the chati-d from the configuration.
parse-mode string Parse node, defaults to Mardown. If empty uses the parse-mode from the configuration.
disable-web-page-preview bool Web Page preview. If empty uses the disable-web-page-preview from the configuration.
disable-notification bool Disables Notification. If empty uses the disable-notification from the configuration.

Example:

kind: telegram

Victorops

Send alert events to a VictorOps instance.

Options:

Name Type Description
routing-key string The routing key of the alert event.

Example:

kind: victorops
options:
    routing-key: ops_team

Match expressions

Alert handlers support match expressions that filter which alert events the handler processes.

A match expression is a TICKscript lambda expression. The data that triggered the alert is available to the match expression, including all fields and tags.

In addition to the data that triggered the alert metadata about the alert is available. This alert metadata is available via various functions.

Name Type Description
level int The alert level of the event, one of ‘0’, ‘1’, ‘2’, or ‘3’ corresponding to ‘OK’, ‘INFO’, ‘WARNING’, and ‘CRITICAL’.
changed bool Indicates whether the alert level changed with this event.
name string Returns the measurement name of the triggering data.
taskName string Returns the task name that generated the alert event.
duration duration Returns the duration of the event in a non OK state.

Additionally the vars OK, INFO, WARNING, and CRITICAL have been defined to correspond with the return value of the level function.

For example to send only critical alerts to a handler, use this match expression:

match: level() == CRITICAL

Examples

Send only changed events to the handler:

match: changed() == TRUE

Send only WARNING and CRITICAL events to the handler:

match: level() >= WARNING

Send events with the tag “host” equal to s001.example.com to the handler:

match: "host" == 's001.example.com'