Skip to content

Integrations

DETAILS: Tier: Free, Premium, Ultimate Offering: GitLab.com, Self-managed, GitLab Dedicated

GitLab can accept alerts from any source via a webhook receiver. Alert notifications can trigger paging for on-call rotations or be used to create incidents.

Integrations list

With at least the Maintainer role, you can view the list of configured alerts integrations by navigating to Settings > Monitor in your project's sidebar menu, and expanding the Alerts section. The list displays the integration name, type, and status (enabled or disabled):

Current Integrations

Configuration

GitLab can receive alerts via a HTTP endpoint that you configure.

Single HTTP Endpoint

Enabling the HTTP Endpoint in a GitLab projects activates it to receive alert payloads in JSON format. You can always customize the payload to your liking.

  1. Sign in to GitLab as a user with the Maintainer role for a project.
  2. Go to Settings > Monitor in your project.
  3. Expand the Alerts section, and in the Select integration type dropdown list, select HTTP Endpoint.
  4. Toggle the Active alert setting. The URL and Authorization Key for the webhook configuration are available in the View credentials tab after you save the integration. You must also input the URL and Authorization Key in your external service.

HTTP Endpoints

DETAILS: Tier: Premium, Ultimate Offering: GitLab.com, Self-managed, GitLab Dedicated

In GitLab Premium, you can create multiple unique HTTP endpoints to receive alerts from any external source in JSON format, and you can customize the payload.

  1. Sign in to GitLab as a user with the Maintainer role for a project.

  2. Go to Settings > Monitor in your project.

  3. Expand the Alerts section.

  4. For each endpoint you want to create:

    1. Select Add new integration.

    2. In the Select integration type dropdown list, select HTTP Endpoint.

    3. Name the integration.

    4. Toggle the Active alert setting. The URL and Authorization Key for the webhook configuration are available in the View credentials tab after you save the integration. You must also input the URL and Authorization Key in your external service.

    5. Optional. To map fields from your monitoring tool's alert to GitLab fields, enter a sample payload and select Parse payload for custom mapping. Valid JSON is required. If you update a sample payload, you must also remap the fields.

    6. Optional. If you provided a valid sample payload, select each value in Payload alert key to map to a GitLab alert key.

    7. To save your integration, select Save Integration. If desired, you can send a test alert from your integration's Send test alert tab after the integration is created.

The new HTTP Endpoint displays in the integrations list. You can edit the integration by selecting the {settings} settings icon on the right side of the integrations list.

Map fields in custom alerts

You can integrate your monitoring tool's alert format with GitLab alerts. To show the correct information in the Alert list and the Alert Details page, map your alert's fields to GitLab fields when you create an HTTP endpoint:

Alert Management List

Customize the alert payload outside of GitLab

For HTTP Endpoints without custom mappings, you can customize the payload by sending the following parameters. All fields are optional. If the incoming alert does not contain a value for the Title field, a default value of New: Alert is applied.

Property Type Description
title String The title of the alert.
description String A high-level summary of the problem.
start_time DateTime The time of the alert. If none is provided, a current time is used.
end_time DateTime The resolution time of the alert. If provided, the alert is resolved.
service String The affected service.
monitoring_tool String The name of the associated monitoring tool.
hosts String or Array One or more hosts, as to where this incident occurred.
severity String The severity of the alert. Case-insensitive. Can be one of: critical, high, medium, low, info, unknown. Defaults to critical if missing or value is not in this list.
fingerprint String or Array The unique identifier of the alert. This can be used to group occurrences of the same alert. When the generic_alert_fingerprinting feature is enabled, the fingerprint is generated automatically based on the payload (excluding the start_time, end_time, and hosts parameters).
gitlab_environment_name String The name of the associated GitLab environment. Required to display alerts on a dashboard.

You can also add custom fields to the alert's payload. The values of extra parameters aren't limited to primitive types (such as strings or numbers), but can be a nested JSON object. For example:

{ "foo": { "bar": { "baz": 42 } } }

NOTE: Ensure your requests are smaller than the payload application limits.

Example request body

Example payload:

{
  "title": "Incident title",
  "description": "Short description of the incident",
  "start_time": "2019-09-12T06:00:55Z",
  "service": "service affected",
  "monitoring_tool": "value",
  "hosts": "value",
  "severity": "high",
  "fingerprint": "d19381d4e8ebca87b55cda6e8eee7385",
  "foo": {
    "bar": {
      "baz": 42
    }
  }
}

Prometheus endpoint

Prerequisites:

  • You must have at least the Maintainer role for the project.
  1. On the left sidebar, select Search or go to and find your project.
  2. Select Settings > Monitor.
  3. Expand the Alerts section, and select Add new integration.
  4. From the Select integration type dropdown list, select Prometheus.
  5. Turn on the Active toggle.
  6. Enter the Prometheus API base URL. You should enter a placeholder URL. The features which use this field are deprecated and scheduled for removal in GitLab 16.0.
  7. Select Save integration.

The URL and authorization key for the webhook configuration are available in the View credentials tab.

Enter the URL and authorization key in your external service. You can also send a test alert from your integration's Send test alert tab.

Add integration credentials to Prometheus Alertmanager

To send Prometheus alert notifications to GitLab, copy the URL and authorization key from your Prometheus integration into the webhook_configs section of the Prometheus Alertmanager configuration:

receivers:
  - name: gitlab
    webhook_configs:
      - http_config:
          authorization:
            type: Bearer
            credentials: 1234567890abdcdefg
        send_resolved: true
        url: http://IP_ADDRESS:PORT/root/manual_prometheus/prometheus/alerts/notify.json
        # Rest of configuration omitted
        # ...

Expected request attributes

Alerts are expected to be formatted for a Prometheus webhook receiver.

Top-level required attributes:

  • alerts
  • commonAnnotations
  • commonLabels
  • externalURL
  • groupKey
  • groupLabels
  • receiver
  • status
  • version

From alerts in the Prometheus payload, a GitLab alert is created for each item in the array. You can alter the nested parameters listed below to configure the GitLab alert.

Attribute Type Required Description
One of annotations/title, annotations/summary, or labels/alertname String Yes The title of the alert.
startsAt DateTime Yes The start time of the alert.
annotations/description String No A high-level summary of the problem.
annotations/gitlab_incident_markdown String No GitLab Flavored Markdown to be appended to any incident created from the alert.
annotations/runbook String No Link to documentation or instructions for how to manage this alert.
endsAt DateTime No The resolution time of the alert.
g0.expr query parameter in generatorUrl String No Query of associated metric.
labels/gitlab_environment_name String No The name of the associated GitLab environment. Required to display alerts on a dashboard.
labels/severity String No Severity of the alert. Should be one of the Prometheus severity options. Defaults to critical if missing or value is not in this list.
status String No Status of the alert in Prometheus. If value is 'resolved', the alert is resolved.
One of annotations/gitlab_y_label, annotations/title, annotations/summary, or labels/alertname String No The Y-Axis label to be used when embedding the metrics for this alert in GitLab Flavored Markdown.

Additional attributes included under annotations are available on the alert details page. Any other attributes are ignored.

Attributes aren't limited to primitive types (such as strings or numbers), but can be a nested JSON object. For example:

{
    "target": {
        "user": {
            "id": 42
        }
    }
}

NOTE: Ensure your requests are smaller than the payload application limits.

Prometheus severity options

Alerts from Prometheus can provide any of the case-insensitive follow values for alert severity:

  • Critical: critical, s1, p1, emergency, fatal
  • High: high, s2, p2, major, page
  • Medium: medium, s3, p3, error, alert
  • Low: low, s4, p4, warn, warning
  • Info: info, s5, p5, debug, information, notice

The severity defaults to critical if the value is missing or not in this list.

Example Prometheus alert

Example alerting rule:

groups:
- name: example
  rules:
  - alert: ServiceDown
    expr: up == 0
    for: 5m
    labels:
      severity: high
    annotations:
      title: "Example title"
      runbook: "http://example.com/my-alert-runbook"
      description: "Service has been down for more than 5 minutes."
      gitlab_y_label: "y-axis label"
      foo:
        bar:
          baz: 42

Example request payload:

{
  "version" : "4",
  "groupKey": null,
  "status": "firing",
  "receiver": "",
  "groupLabels": {},
  "commonLabels": {},
  "commonAnnotations": {},
  "externalURL": "",
  "alerts": [{
    "startsAt": "2022-010-30T11:22:40Z",
    "generatorURL": "http://host?g0.expr=up",
    "endsAt": null,
    "status": "firing",
    "labels": {
      "gitlab_environment_name": "production",
      "severity": "high"
    },
    "annotations": {
      "title": "Example title",
      "runbook": "http://example.com/my-alert-runbook",
      "description": "Service has been down for more than 5 minutes.",
      "gitlab_y_label": "y-axis label",
      "foo": {
        "bar": {
          "baz": 42
        }
      }
    }
  }]
}

Authorization

The following authorization methods are accepted:

  • Bearer authorization header
  • Basic authentication

The <authorization_key> and <url> values can be found when configuring an alert integration.

Bearer authorization header

The authorization key can be used as the Bearer token:

curl --request POST \
  --data '{"title": "Incident title"}' \
  --header "Authorization: Bearer <authorization_key>" \
  --header "Content-Type: application/json" \
  <url>

Basic authentication

The authorization key can be used as the password. The username is left blank:

  • username: <blank>
  • password: <authorization_key>
curl --request POST \
  --data '{"title": "Incident title"}' \
  --header "Authorization: Basic <base_64_encoded_credentials>" \
  --header "Content-Type: application/json" \
  <url>

Basic authentication can also be used with credentials directly in the URL:

curl --request POST \
  --data '{"title": "Incident title"}' \
  --header "Content-Type: application/json" \
  <username:password@url>

WARNING: Using your authorization key in the URL is insecure, as it's visible in server logs. We recommend using one of the above header options if your tooling supports it.

Response body

The JSON response body contains a list of any alerts created within the request:

[
  {
    "iid": 1,
    "title": "Incident title"
  },
  {
    "iid": 2,
    "title": "Second Incident title"
  }
]

Successful responses return a 200 response code.

Triggering test alerts

After a project maintainer or owner configures an integration, you can trigger a test alert to confirm your integration works properly.

  1. Sign in as a user with at least the Developer role.
  2. Go to Settings > Monitor in your project.
  3. Select Alerts to expand the section.
  4. Select the {settings} settings icon on the right side of the integration in the list.
  5. Select the Send test alert tab to open it.
  6. Enter a test payload in the payload field (valid JSON is required).
  7. Select Send.

GitLab displays an error or success message, depending on the outcome of your test.

Automatic grouping of identical alerts

DETAILS: Tier: Premium, Ultimate Offering: GitLab.com, Self-managed, GitLab Dedicated

GitLab groups alerts based on their payload. When an incoming alert contains the same payload as another alert (excluding the start_time and hosts attributes), GitLab groups these alerts together and displays a counter on the Alert Management List and details pages.

If the existing alert is already resolved, GitLab creates a new alert instead.

Alert Management List

Recovery alerts

The alert in GitLab is automatically resolved when an HTTP Endpoint receives a payload with the end time of the alert set. For HTTP Endpoints without custom mappings, the expected field is end_time. With custom mappings, you can select the expected field.

GitLab determines which alert to resolve based on the fingerprint value that can be provided as part of the payload. For more information on alert properties and mappings, see Customize the alert payload outside of GitLab.

You can also configure the associated incident to be closed automatically when the alert resolves.

Link to your Opsgenie Alerts

DETAILS: Tier: Premium, Ultimate Offering: GitLab.com, Self-managed, GitLab Dedicated

WARNING: We are building deeper integration with Opsgenie and other alerting tools through HTTP endpoint integrations so you can see alerts in the GitLab interface.

You can monitor alerts using a GitLab integration with Opsgenie.

If you enable the Opsgenie integration, you can't have other GitLab alert services active at the same time.

To enable Opsgenie integration:

  1. Sign in as a user with the Maintainer or Owner role.
  2. Go to Monitor > Alerts.
  3. In the Integrations select box, select Opsgenie.
  4. Select the Active toggle.
  5. In the API URL field, enter the base URL for your Opsgenie integration, such as https://app.opsgenie.com/alert/list.
  6. Select Save changes.

After you enable the integration, go to the Alerts page at Monitor > Alerts, and then select View alerts in Opsgenie.