View and manage alerts in Elastic Observability

You can track and manage alerts for your applications and SLOs from the Alerts page. You can filter this view by alert status or time period, or search for specific alerts using KQL. Manage your alerts by adding them to cases or viewing them within the respective UIs.

×

To help you get started with your analysis faster, use the KQL bar to create structured queries using Kibana Query Language.

You can use the time filter to define a specific date and time range. By default, this filter is set to search for the last 15 minutes.

You can also filter by alert status using the buttons below the KQL bar. By default, this filter is set to Show all alerts, but you can filter to show only active, recovered or untracked alerts.

There are a few ways to inspect the details for a specific alert.

From the Alerts table, you can click on a specific alert to open the alert detail flyout to view a summary of the alert without leaving the page. There you’ll see the current status of the alert, its duration, and when it was last updated. To help you determine what caused the alert, you can view the expected and actual threshold values, and the rule that produced the alert.

×

To further inspect the rule:

• From the alert detail flyout, click View rule details.
• From the Alerts table, click the icon and select View rule details.

To view the alert in the app that triggered it:

• From the alert detail flyout, click View in app.
• From the Alerts table, click the icon.

Check related alerts to find other alerts that might be related to the same incident. You can add these alerts to a case and investigate them as a group instead of analyzing them individually.

To find related alerts, go to the Related alerts tab from an alert's details page. Within the table, alerts are ordered from most to least relevant. To only view alerts that were created around the same time as the current alert (+/- 30 minutes), apply the Triggered around the same time filter.

The relevancy of alerts is determined by how closely they match the current alert and other similiarites that they might share:

1. Alerts in the space are filtered down to only include alerts that were created about one day before or after the current alert.
2. Data from the new subset of alerts is compared against the current alert to identify matching values and similarities. Data such as the time at which alerts were generated or recovered, tags added to the alerts, group values, and more are evaluated.
3. Alerts are scored based on how closely they match the current alert. Alerts with a score above a certain threshold are considered relevant and are included in the list of related alerts.

There are four common alert statuses:

active

The conditions for the rule are met. If the rule has actions, Kibana generates notifications based on the actions' notification settings.

flapping

The alert switched repeatedly between active and recovered states. If actions are configured to run when its status changes, they are suppressed. Refer to Configure alert flapping to learn more about configuring alert flapping for rules.

recovered

The conditions for the rule are no longer met. If the rule has recovery actions, Kibana generates notifications based on the actions' notification settings. Recovery actions only run if the rule's conditions aren't met during the current rule execution, but were in the previous one.

An active alert changes to recovered if the conditions for the rule that generated it are no longer met.

A flapping alert changes to recovered when the rule's conditions are unmet for a specific number of consecutive runs. This number is determined by the Alert status change threshold setting, which you can configure under the Alert flapping detection settings.

For example, if the threshold requires an alert to change status at least 6 times in the last 10 runs to be considered flapping, then to recover, the rule's conditions must remain unmet for 6 consecutive runs. If the rule's conditions are met at any point during this recovery period, the count of consecutive unmet runs will reset, requiring the alert to remain unmet for an additional 6 consecutive runs to finally be reported as recovered.

Once a flapping alert is recovered, it cannot be changed to flapping again. Only new alerts with repeated status changes are candidates for the flapping status.

untracked

The rule is disabled, or you’ve marked the alert as untracked. To mark the alert as untracked, go to the Alerts table, click the action menu ( ) to expand the More actions menu, and click Mark as untracked. When an alert is marked as untracked, actions are no longer generated and the alert's status can no longer be changed. You can choose to move active alerts to this state when you disable or delete rules.

If an alert is active or flapping, you can mute it to temporarily suppress future actions. While muted, the alert's status will continue to update but rule actions won't run. All future alerts with the same alert ID will also be muted. You can mute alerts in the following ways:

Use alert tags to organize related alerts into categories that you can filter and group. For example, use the Production alert tag to label a group of alerts as notifications from your production environment. Then, to find alerts with the Production tag, enter the kibana.alert.workflow_tags : "Production" query into the Alert's table KQL bar.

To apply or remove alert tags on individual alerts:

1. Go to the Alerts table, click the More actions menu ( ) in an alert’s row, then click Edit tags.

2. In the flyout, do one of the following:

   • Apply a new tag: Enter a new tag into the search bar, then select the Add tag name as a tag or click enter on your keyboard to apply your changes.

   • Remove existing tags: Click the tag that you want to remove. To remove all tags from the alert, click Select none.

     Important

     Removing tags from an alert permanently deletes them.

3. Click Save selection to apply your changes to the alert.

To apply or remove alert tags on multiple alerts, select the alerts you want to change, then click Selected x alerts at the upper-left above the table. Click Edit alert tags, select or unselect tags, then click Save selection.

Use the toolbar buttons in the upper-left of the alerts table to customize the columns you want displayed:

• Columns: Reorder the columns.
• x fields sorted: Sort the table by one or more columns.
• Fields: Select the fields to display in the table.

For example, click Fields and choose the Maintenance Windows field. If an alert was affected by a maintenance window, its identifier appears in the new column. For more information about their impact on alert notifications, refer to Maintenance windows.

You can also use the toolbar buttons in the upper-right to customize the display options or view the table in full-screen mode.

From the Alerts table, you can add one or more alerts to a case. Click the icon to add the alert to a new or existing case. You can add an unlimited amount of alerts from any rule type.

To add an alert to a new case:

1. Select Add to new case.
2. Enter a case name, add relevant tags, and include a case description.
3. Under External incident management system, select a connector. If you’ve previously added one, that connector displays as the default selection. Otherwise, the default setting is No connector selected.
4. After you’ve completed all of the required fields, click Create case. A notification message confirms you successfully created the case. To view the case details, click the notification link or go to the Cases page.

To add an alert to an existing case:

1. Select Add to existing case.
2. Select the case where you will attach the alert. A confirmation message displays.

Manage the size of alert indices in your space by clearing out alerts that are older or infrequently accessed. You can do this by running an alert cleanup task, which deletes alerts according to the criteria that you define.