Creating Chronograf alert rules

Chronograf provides a user interface for Kapacitor, InfluxData’s processing framework for creating alerts, running ETL jobs, and detecting anomalies in your data. Chronograf alert rules correspond to Kapacitor tasks that are designed specifically to trigger alerts whenever data stream values rise above or fall below designated thresholds. Common alerting use cases that can be managed using Chronograf include:

  • Thresholds with static ceilings, floors, and ranges.
  • Relative thresholds based on unit or percentage changes.
  • Deadman switches.

Complex alerts and other tasks must be defined directly in Kapacitor, but can be used within Chronograf.

Follow this guide to create a Chronograf alert rule that sends an alert message to an existing Slack channel whenever your idle CPU usage crosses the 80% threshold.

Requirements

Getting started with Chronograf guide offers step-by-step instructions for each of the requirements here.

Configuring Chronograf alert rules

Navigate to the Rule Configuration page by selecting the Alerting page and then clicking Create Rule in the top right corner.

Navigate to Rule Configuration

The Rule Configuration page is used to create and edit your Chronogra alert rules. The steps below guide you through the process of creating a Chronograf alert rule.

Empty Rule Configuration

Step 1: Name the alert rule

Click Untitled Rule in the top left corner and name the rule.

For this example, the rule is named Idle CPU Usage:

Name your rule

Step 2: Select the time series data

Choose the time series data that you want the Chronograf alert rule to use. Navigate through the Databases, Measurements, Fields, and Tags tabs to select the relevant data.

In this example, we select the telegraf database and the autogen retention policy, the cpu measurement, the usage_idle field, and no tags. The result is the InfluxQL query in the image below. Notice that Chronograf automatically sets a time range in the WHERE clause. Don’t modify that for now. The time range is discussed in step four.

Select your data

Step 3: Select the alert type

Choose from three alert types in the Rule Conditions section of the Rule Configuration page. The three alert types are:

  • Threshold: Alert if the data cross a boundary
  • Relative: Alert if the data change relative to the data in a different time range
  • Deadman: Alert if InfluxDB receives no relevant data for the specified time duration

For this example, select the Threshold alert type.

Step 4: Define the rule condition

Define the threshold condition:

Create a condition

Moving across the inputs from right to left:

  • usage_idle: The field key specified in the Select a Time Series section.
  • Less than: The condition type. Chronograf supports several condition types.
  • 80: The threshold number. The system sends an alert when the usage_idle data cross that boundary.

The graph shows a preview of the relevant data and the threshold number. By default, the graph shows data from the past 15 minutes. The time range selector in the top right corner adjusts the graph’s time range. Use this feature when determining a reasonable threshold number based on your data.

Note: We set the threshold number to 80 for demonstration purposes. On our machine, setting the threshold for idle CPU usage to a high number ensures that we’ll be able to see the alert in action. In practice, you’d set the threshold number to better match the patterns in your data and your alert needs.

Step 5: Select the event handler and configure the alert message

The Alert Message section on the Rule Configuration page determines where the system sends the alert (the event handler) and the text that accompanies the alert (the alert message). Chronograf supports several event handlers. Here, we choose to send alerts to Slack, specifically to the existing #chronocats channel.

The alert message is the text that accompanies an alert. In this example, the alert message is Your idle CPU usage is {{.Level}} at {{ index .Fields "value" }}. 😸. {{.Level}} is a template that evaluates to CRITICAL when the usage_idle data initially dip below 80% and OK when the usage_idle data first return to 80% or above. The {{ index .Fields "value" }} template prints the relevant field value that triggered the alert.

Specify event handler and alert message

Note: There’s no need to include a Slack channel in the Alert Message section if you specified a default channel in the initial Slack configuration. If you did not include a default channel in the initial configuration or if you’d like to send alerts to a non-default channel, specify an alternative Slack channel in this section.

Step 6: Save the alert rule

Click Save Rule in the top right corner and navigate to the Alert Rule page to see your rule. Notice that you can easily enable and disable the rule by toggling the checkbox in the Enabled column.

See the alert rule

Next, move on to the section below to experience your alert rule in action.

Viewing alerts in practice

Step 1: Create some load on your system

The purpose of this step is to generate enough load on your system to trigger an alert. More specifically, your idle CPU usage must dip below 80%. On the machine that’s running Telegraf, enter the following command in the terminal to start some while loops:

while true; do i=0; done

Let it run for a few seconds or minutes before terminating it. On most systems, kill the script by using Ctrl+C.

Step 2: Visit Slack

Go to the Slack channel that you specified in the previous section. In this example, it’s the #chronocats channel.

Assuming the first step was successful, #chronograf should reveal at least two alert messages:

  • The first alert message indicates that your idle CPU usage was CRITICAL, that is, it dipped below 80%. The specific field value that triggered the alert is 69.59999999998138.
  • The second alert message indicates that your idle CPU usage returned to an OK level of 80% or above. The specific field value that triggered the alert is 99.0981963931105.

See the alerts

That’s it! You’ve successfully used Chronograf to configure an alert rule to monitor your idle CPU usage.

This documentation is open source. See a typo? Please, open an issue.


Need help getting up and running? Get Support