---
title: "Create an alert"
description: "Create an alert that names a service, a metric, a condition, a severity, and the monitoring integrations to notify."
content_type: "how-to"
product: "alerts"
platform: "TIR"
url: "/tir/alerts/how-to/create-an-alert/"
topic: ["create-and-use-alerts"]
relationships: {"requires":["/tir/alerts/how-to/create-a-monitoring-integration/"]}
---

# Create an alert

Create an alert to set when TIR notifies you and who it notifies. The alert starts working when you
[attach it to a running resource](/tir/alerts/how-to/attach-an-alert-to-a-resource/).

## Before you begin

You need a role with Create permission on alerts, and at least one monitoring
integration. [Create one](/tir/alerts/how-to/create-a-monitoring-integration/) first if you need
to, or use the link in the form to create one there.

## Open the form

In TIR, open your project. In the side menu, under **PLATFORM SERVICES**, select **Alert
Management**, then **Create Alert**. You can also create an alert from a resource's alert panel:
select the link in the **Alerts** panel. The form then opens with the service already set to that
resource.

## Fill in the form

All fields are required.

| Field                      | What to enter                                                                                                                                     |
| -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Alert name**             | A generated name; replace it with a descriptive one such as `prod-gpu-temp-critical`.                                                             |
| **Service for alert**      | The kind of resource: **Instance**, **VM**, **Inference**, **Training Cluster**, or **Load Balancer**. It stays fixed after you create the alert. |
| **Monitoring Integration** | One or more integrations to notify, such as `ops-slack-webhook`. The field shows how many you picked.                                             |
| **Alert trigger type**     | The metric. The list depends on the service; see [Alert metrics and conditions](/tir/alerts/concepts/alert-metrics-and-conditions/).              |
| **Severity**               | **Critical**, **Warning**, or **Info**.                                                                                                           |
| **Operator**               | How to compare: **Greater Than**, **Less Than**, **Greater Than or Equal To**, **Less Than or Equal To**, **Equal To**, or **Not Equal To**.      |
| **Threshold value**        | A number. The unit shows in the label, such as (%) or (C). A percentage metric takes 0 to 100.                                                    |

**Load Balancer** is listed in Delhi and Chennai. From the **Alerts** tab of a TCP load balancer,
**Alert trigger type** lists only **Healthy Backends %** and **Active Connections**.

An alert name starts with a letter, uses letters, numbers, hyphens, and underscores, and ends with a
letter or number. It can be up to 255 characters and must be unique in the project.

For **XID Error**, **Operator** and **Threshold value** are set for you to **Not Equal To** and 0:
the alert fires when the GPU reports an XID error. For **Nodes Down**, the alert fires whenever more
than 0 nodes are down, whatever you enter.

## Create it

Select **Create**. The alert is in the list, ready to attach.

## Find the alert in the list

In **Alert Management**, the alert is listed with its resource type, severity, metric, operator,
threshold, and number of integrations. Its **Status** is a dash until you attach it. To start the alert watching, [attach it to a
resource](/tir/alerts/how-to/attach-an-alert-to-a-resource/).