Create an alert

Create an alert that watches one measurement, such as memory in use or free disk space. It can cover as many resources of one service as you choose, and it takes a single row on Manage Alerts. When the measurement crosses the threshold you set, everyone in the alert groups you pick gets an email.

To change an alert you already have, see Edit, pause, or delete an alert. To set an alert while you work on one resource, use that resource’s own Alerts tab; Event Monitoring links each product’s steps.

Before you begin

You need:

• At least one running node, load balancer, DBaaS instance, or Kubernetes cluster to watch. The form offers only running resources.
• Create permission in Event Monitoring for the project. With it, MyAccount shows Create Alert. An Admin can give it to you.

Alerts belong to the project and region selected in MyAccount, Delhi or Chennai. Select the project and region of the resources you want to watch before you start.

You also need an alert group, the list of people the alert notifies. You can create one from inside the form.

Open the alert form

1. In the side menu, click Event Monitoring. The Monitoring Dashboard opens.
2. Click Manage Alerts. The list of your alerts opens.
3. Click Create Alert. The Create Alert form opens.

Choose the resources to watch

1. In Service for alert, choose Node, Load Balancer, DBaaS, or Kubernetes.
2. For Kubernetes, in Associated Resource, first choose the cluster. The next field then lists that cluster’s nodes.
3. In Associated Resources, choose the resources the alert covers, or choose All to pick every resource in the list. Only running resources of that service are listed.

When nothing of that service is running in this project and region, Associated Resources shows No resources found. Create needs at least one resource, so start one, or choose another service.

Choose the resources before the trigger type. If you change the resources afterwards, the trigger type is cleared and you choose it again.

Choose what to measure

In Alert trigger type, choose the measurement. The list depends on the service.

For Node:

• Processor Load: how busy the machine’s vCPUs are, as a percentage of its vCPU count.
• Consumed Memory: the share of memory in use, in percent.
• Free Disk Space: the share of the disk that is still free, in percent.
• Volume Read Ops and Volume Write Ops: reads or writes on one disk. These two appear when a single node is selected.

For Load Balancer and Kubernetes, the list is Processor Load, Consumed Memory, and Free Disk Space.

For DBaaS, an alert watches the DBaaS instance and each of its Read Replicas. The list is:

• CPU Load Average: how busy the vCPUs are, as a percentage of their count.
• CPU Utilization: the share of CPU in use, in percent.
• Free Disk Space: the share of the disk that is still free, in percent.
• Free Memory, Slow Query Rate, and Query per second, each with a threshold from 10 to

To set one from the DBaaS instance itself, see Monitor a DBaaS instance and set alerts.

To check that a website on a node still answers, use a Webcheck alert, which you set up from the node’s own Alerts tab. See Set up alerts on a node.

Pick the disk

For Free Disk Space, Volume Read Ops, and Volume Write Ops, a Volume field appears. Choose Node Volume(Default) for the machine’s own disk. With a single node selected, you can choose one of its attached volumes instead.

Before you choose an attached volume for a Free Disk Space alert, mount it on the node and turn on its monitoring. See Monitor a volume.

Set the condition

1. In Operator, choose Greater than, Less than, Equal to, or Not equal to. Greater than tells you when a value climbs, which suits Processor Load, Consumed Memory, CPU Load Average, CPU Utilization, and the volume operations. Free Disk Space counts what is left, so to hear about a disk filling up, choose Less than.
2. In Threshold value (%), enter a number from 10 to 100. For Volume Read Ops and Volume Write Ops, the field has no percent sign and takes a number from 1 to 75,000. If you enter a number outside the range, the form shows the allowed range until you correct it.
3. In Severity, choose Not classified, Information, Warning, Average, High, or Disaster. The form starts at High. Severity labels the problem so your team can tell how urgent it is; the operator and threshold decide when the alert fires. The Monitoring Dashboard counts open problems by severity.

What a Processor Load percentage means

A Processor Load threshold is a share of the machine’s vCPU count, measured against the load average. The load average is the average number of processes running or waiting for a CPU. On a node with two vCPUs, 100 means a load average of two. With Greater than and 80 on a node with four vCPUs, the alert fires when the five-minute load average stays above 3.2 for five minutes.

The CPU Load Average trigger for a DBaaS instance works the same way, on the one-minute load average.

Choose who is notified

In Alert Group(s), choose one or more alert groups, or All for every group. You can search the list by name. The list shows the alert groups of the current project, and an alert can notify only those.

To make a new group without leaving the form, click the plus icon beside Alert Group(s). Create an alert group from the alert form has the steps.

When the project has no alert groups yet, the field reads No group Available. Create needs at least one alert group, so make one first.

When the alert fires, every member of each group you chose gets an email. Each problem sends one, when it starts. How alerts reach people covers the notifications by email and on WhatsApp.

Check the summary and create the alert

Once every field is filled in, an Alert Summary sentence appears under the form. Read it to catch a wrong operator or threshold before you save.

Click Create. MyAccount shows Alert created successfully. and opens Manage Alerts, where the new alert is listed. Its Alert Type reads Trigger : followed by the measurement, such as Memory. Its Alert Condition reads like “Alert when Memory is greater than 80 %”. To leave the form without saving, click Cancel.

For help with any message, see Troubleshooting Event Monitoring.

When the alert’s condition holds on one of its resources, a problem appears on the Monitoring Dashboard. To find it and open the resource’s graphs, see Read the monitoring dashboard. To raise the threshold later, or to pause the notifications during maintenance, see Edit, pause, or delete an alert.