GridGain Developers Hub

Configuring Alerts for Apache Ignite

Former Head of Developer Relations at GridGain
Apache Ignite Committer and PMC Member

An alert is a trigger that ensures that you are notified when a specified cluster event occurs. By configuring alerts, you automate the monitoring of your Apache Ignite clusters. For example, you might want to be notified when off-heap memory or disk usage goes beyond 20GB or when the number of cluster nodes is reduced to eight.

GridGain Nebula comes with then built-in alerting capabilities, window="_blank" that enable you to configure an alert for any Ignite metric. In this part of the tutorial, you learn how to configure a couple of alerts. At the end of the process, your alerting screen should look like the following:

Ignite Alerting

Configure a Custom Notification Channel

To ensure that you are notified as soon as an alert fires, Nebula requires that you set up at least one notification channel. When this tutorial was written, you could configure an email or an SMS notification channel.

Open the Alerting screen, window="_blank" and create a notification channel:

Ignite Alerting Channel
  • Name: Specify a channel name; for now, use My Email Server.

  • Send with: Specify a real mail server. For Control Center installations, you can skip this configuration step for now and provide the mail server settings later.

  • Send to: Specify an email address to receive notifications, such as rescue_team@company.com.

Create a Memory-Usage Alert

In the second part of the tutorial, you configured the Storage Usage dashboard, which displays various memory and disk-usage metrics. Let’s go further and set up an alert for one of the metrics.

Create an alert that is triggered when off-heap memory usage exceeds 500MB on any node:

Ignite Memory Usage Alert
  • Name: Specify a custom name for the alert, such as Off-Heap Memory Usage Threshold.

  • Metric Name: In the metrics dialog, select the io.dataregion.default.PhysicalMemorySize metric.

  • Condition and Value: Set the condition to trigger the alert when off-heap usage is greater than 500000000 bytes (500MB).

  • Grace Period: Set the grace period such that the alert fires only if the condition lasts for 10000ms (10 secs).

  • Notification Channels: Select the My Email Server channel that you configured previously.

Set Up a Node-Count Alert

Number of server nodes can be a critical parameter for Ignite clusters in production. If a cluster contains too few nodes, the cluster’s performance or availability can be negatively affected. Nebula enables you to configure a node-count alert that monitors the cluster’s size and notifies you if size poses a risk.

Configure an alert that is triggered every time the cluster’s size goes below two server nodes:

Ignite Nodes-Count Alert
  • Name: Use Cluster Nodes Threshold as the name.

  • Metric Name: In the metrics dialog, select the cluster.ServerNodesCloud metric.

  • Condition and Value: Set the condition such that the alert is triggered if the number of nodes is less than two.

  • Grace Period: Set the grace period to 0 to ensure that the alert is fired without delay.

  • Notification Channels: Select the My Email Server channel that you configured previously.

Review the Final List of Alerts

Now, you should have two alerts, which are defined as shown in the following screenshot:

Ignite Alerting

What’s Next

Complete the next part of the tutorial to learn how to use tracing to troubleshoot bottlenecks and hot spots: