Setting up Real-Time Alerts with Alert Center

This guide explains how to set up metric, log, and event-based alerts in KakaoCloud's Alert Center.

Basic Information

• Estimated time: 30 minutes
  
• Recommended user environment
  • Recommended OS: MacOS, Ubuntu
    
  • Region: kr-central-2
    

Before you start

KakaoCloud's Alert Center allows you to configure alert policies and notification channels to receive alerts for service events and resource failures. Setting up timely notifications helps you detect and resolve issues quickly.

About this scenario

This tutorial provides step-by-step instructions for configuring metric, log, and event-based alerts in Alert Center, allowing you to receive real-time notifications.

• Notification channel setup: Create and configure channels to receive alerts in Alert Center.
• Metric-based alerts: Set up alerts by simulating a CPU load test.
• Log-based alerts: Create a log-based alert by generating error logs.
• Event-based alerts: Set up an event-based alert triggered by instance restarts.

Before you start

1. Install monitoring agent

To monitor compute resources on your instance, you must install a monitoring agent. For detailed instructions, refer to the BCS Instance Monitoring tutorial.

2. Enter contact information

Ensure you have entered your email address and phone number correctly to receive alerts. For setup instructions, refer to the Modify account information guide.

3. Create notification channels in Alert Center

In KakaoCloud's Alert Center, you can Create notification channels to receive alerts for compute resource risk levels.

1. Go to KakaoCloud Console > Management > Alert Center > Notification Channels.

2. Click the [Create notification channel] button.

3. In the notification channel creation tab, enter the required information, select the channel type, and click [Create].

   Type	Description
   Default	Sends alerts via email/SMS/KakaoTalk
   Slack/Webhook	Sends alerts to a specified Slack channel or webhook

info

• The log file path for the monitoring agent used in this tutorial is /var/log/messages, and the notification channel type is Default (email).

Getting started

The following instructions explain how to configure various alert types (metric, log, event) in Alert Center.

1. Metric-based alert

Perform a CPU load test (stress) on an instance with the monitoring agent installed to confirm the metric-based alert.

Step 1. Create alert policy

Use KakaoCloud's Alert Center to configure alert policies based on compute resource risk levels.

1. In Alert Center > Notification Channel, click [Create alert policy].

2. Set the necessary alert conditions and click [Create].

   Item	Value
   Condition type	Metric
   Service	Beyond Compute Service
   Condition	- Metric: CPU usage - Resource: instance (select monitoring instance) - Threshold: over 70% - Duration: 1 minute - Severity: Critical
   Notification Channel	Select the preconfigured notification channel
   Alert policy name	tutorial-alert-metric
   Enabled	Yes

Step 2. Test CPU load and confirm alert

On the instance with the monitoring agent installed, use the stress tool to generate CPU load and verify the alert is received.

1. SSH into the instance with the monitoring agent installed.

2. Install the stress tool.

   sudo apt-get update
   sudo apt-get install stress

3. Generate CPU load for 2 minutes.

   stress --cpu 2 --timeout 2m
   
   stress: info: [2221] dispatching hogs: 2 cpu, 0 io, 0 vm, 0 hdd # Pausing for 2 minutes is expected behavior
   stress: info: [2221] successful run completed in 120s # Output after 2 minutes

4. Confirm that the metric-based alert for CPU load was sent to the registered email address from KakaoCloud.

   Check Alert

2. Log-based alert

This section explains how to set up a log-based alert triggered by the occurrence of error messages in the log files of an instance with the monitoring agent installed.

Step 1. Create alert policy

You can configure log-based alert policies in KakaoCloud's Alert Center.

1. In Alert Center > Notification Channel, click [Create alert policy].

2. Set the necessary alert conditions and click [Create].

   Item	Value
   Condition type	Log
   Service	Beyond Compute Service
   Condition	- Keyword: error ⚠️ After entering the keyword, click + to add it.
   Notification Channel	Select the preconfigured notification channel
   Alert policy name	tutorial-alert-log
   Enabled	Yes

Step 2. Check error alert

Run a test script on the instance with the monitoring agent installed to verify the log-based alert.

1. SSH into the instance with the monitoring agent installed.

2. Create and run the error test script.

   # Create script file
   touch script.sh
   
   # Open the file
   sudo vi script.sh

   Enter the following script in script.sh and save it.

   script.sh

   #!/bin/bash
   
   # Log file path set during monitoring agent installation
   log_file="/var/log/messages"
   
   echo "Error: Something went wrong" >&2
   
   # Write to log file with administrator privileges
   sudo bash -c "echo $(date): Custom error occurred >> $log_file"

   Grant execute permissions and run the script.

   # Grant execute permissions to the script
   sudo chmod +x script.sh
   
   # Run the script
   ./script.sh
   
   # Output when the script is executed
   Error: Something went wrong

3. Confirm that the log-based alert for the error keyword was sent to the registered email address from KakaoCloud.

   Check Alert

3. Event-based alert

This section explains how to set up an event-based alert triggered by instance restart events in KakaoCloud's Alert Center.

Step 1. Create alert policy

You can configure event-based alert policies in KakaoCloud's Alert Center.

1. In Alert Center > Notification Channel, click [Create alert policy].

2. Set the necessary alert conditions and click [Create].

   Item	Value
   Condition type	Event
   Service	Virtual Machine
   Condition	- Event: Instance Restart
   Notification Channel	Select the preconfigured notification channel
   Alert policy name	tutorial-alert-event
   Enabled	Yes

Step 2. Verify alerts related to the restart event

Restart the virtual machine instance and verify that the event-based alert was sent.

Check Alert