> ## Documentation Index
> Fetch the complete documentation index at: https://help.dsalta.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AWS Elasticsearch cluster CPU utilization should be monitored

> Checks that Elasticsearch CPU utilization is monitored.

## What this check does

Checks for CloudWatch alarms on CPUUtilization.

| | |
| - | - |
| **Severity** | Medium |
| **Category** | `elasticsearch` |
| **Data used** | ES CPU alarm status |
| **Permissions required** | `es:ListDomainNames`, `cloudwatch:DescribeAlarms` |

**Passes when:** All Elasticsearch domains have CPU monitoring.

## Why this matters

CPU monitoring detects performance issues.

## How to fix

**Create a CloudWatch alarm on OpenSearch CPU utilization**

1. Sign in to the [AWS Console](https://console.aws.amazon.com/) and navigate to **CloudWatch → Alarms → All alarms**, then click **Create alarm**.
2. Click **Select metric**, open the **ES** namespace, and choose the **CPUUtilization** metric for the affected OpenSearch/Elasticsearch domain, then click **Select metric**.
3. Under **Conditions**, set the alarm to trigger when **CPUUtilization** is **Greater** than **80** percent for 15 minutes.
4. Click **Next** and, under **Notification**, select or create an **SNS topic** that alerts your team (email, Slack, or PagerDuty).
5. Click **Next**, name the alarm (e.g. `OpenSearch-CPUUtilization-High`), and click **Create alarm**.

Once resolved, DSALTA records a **200** (compliant) result on this test's **Source Data** tab at the next scheduled run.

## Frequently asked questions

> [accordion] How often does this check run?
>
> Once a day, at 02:00 America/New\_York, while the integration is connected. A run is skipped entirely if the integration is not in a **connected** state, or if the test has been deactivated.
>
> To see a fix reflected sooner, open the test from **Compliance → Tests** and click **Run** in the **Run Test** row. You need Full Access to run a test on demand.

> [accordion] What does the status actually tell me?
>
> The **Status** column answers "did the check run?", not "did I pass". **Completed** means DSALTA reached your system and finished evaluating — a check that ran and *found a problem* is still Completed.
>
> To see what it found, open the test and read the result code on the **Source Data** tab. **200** means nothing to fix; **204** means the check did not apply to your environment; **207** means it ran and found a compliance gap you need to act on. See [Understanding test results](https://help.dsalta.com/guides/compliance/test-results.md).

> [accordion] The check shows Failed. Does that mean I am non-compliant?
>
> Usually not. On an integration-powered check, **Failed** means DSALTA was *blocked* and could not evaluate it — result code **403** (a missing permission) or **428** (a setting DSALTA still needs from you). Code **500** means the run failed on DSALTA's side.
>
> A genuine compliance gap does not turn the list red — it runs to completion and shows **207**.

> [accordion] What if this check does not apply to us?
>
> Deactivate it: open the test from **Compliance → Tests**, then use the **⋮** menu at the top right of the panel and choose **Deactivate Test**. It applies immediately, with no confirmation step.
>
> Write your justification in the **Comments** tab **before** you deactivate. Once deactivated, every other tab is grayed out, so Comments is the only place an auditor can read why — see [Test detail](https://help.dsalta.com/guides/compliance/test-detail.md).

> [accordion] Does DSALTA change my configuration?
>
> No. DSALTA only reads. It never creates, modifies, or deletes resources, and every remediation step is performed by your team directly in the third-party product.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.