Once enabled, GitLab will automatically detect metrics from known services in the [metric library](#monitoring-cicd-environments). You are also able to [add your own metrics](#adding-additional-metrics-premium) as well.
Prometheus is deployed into the `gitlab-managed-apps` namespace, using the [official Helm chart](https://github.com/helm/charts/tree/master/stable/prometheus). Prometheus is only accessible within the cluster, with GitLab communicating through the [Kubernetes API](https://kubernetes.io/docs/concepts/overview/kubernetes-api/).
The Prometheus server will [automatically detect and monitor](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#kubernetes_sd_config) nodes, pods, and endpoints. To configure a resource to be monitored by Prometheus, simply set the following [Kubernetes annotations](https://kubernetes.io/docs/concepts/overview/working-with-objects/annotations/):
CPU and Memory consumption is monitored, but requires [naming conventions](prometheus_library/kubernetes.md#specifying-the-environment) in order to determine the environment. If you are using [Auto DevOps](../../../topics/autodevops/), this is handled automatically.
The [NGINX Ingress](../clusters/index.md#installing-applications) that is deployed by GitLab to clusters, is automatically annotated for monitoring providing key response metrics: latency, throughput, and error rates.
GitLab will automatically scan the Prometheus server for metrics from known servers like Kubernetes and NGINX, and attempt to identify individual environment. The supported metrics and scan process is detailed in our [Prometheus Metrics Library documentation](prometheus_library/index.md).
Custom metrics can be monitored by adding them on the monitoring dashboard page. Once saved, they will be displayed on the environment performance dashboard provided that either:
- A [connected Kubernetes cluster](../clusters/add_remove_clusters.md) with the environment scope of `*` is used and [Prometheus installed on the cluster](#enabling-prometheus-integration)
- **Y-axis label**: Y axis title to display on the dashboard.
- **Unit label**: Query units, for example `req / sec`. Shown next to the value.
Multiple metrics can be displayed on the same chart if the fields **Name**, **Type**, and **Y-axis label** match between metrics. For example, a metric with **Name**`Requests Rate`, **Type**`Business`, and **Y-axis label**`rec / sec` would display on the same chart as a second metric with the same values. A **Legend label** is suggested if this feature used.
GitLab supports a limited set of [CI variables](../../../ci/variables/README.md) in the Prometheus query. This is particularly useful for identifying a specific environment, for example with `ci_environment_slug`. The supported variables are:
There are 2 methods to specify a variable in a query or dashboard:
1. Variables can be specified using the [Liquid template format](https://help.shopify.com/en/themes/liquid/basics), for example `{{ci_environment_slug}}` ([added](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/20793) in GitLab 12.6).
1. You can also enclose it in quotation marks with curly braces with a leading percent, for example `"%{ci_environment_slug}"`. This method is deprecated though and support will be [removed in the next major release](https://gitlab.com/gitlab-org/gitlab/issues/37990).
#### Editing additional metrics from the dashboard
> [Introduced](https://gitlab.com/gitlab-org/gitlab/issues/208976) in GitLab 12.9.
You can edit existing additional custom metrics by clicking the **{ellipsis_v}** **More actions** dropdown and selecting **Edit metric**.
The following tables outline the details of expected properties.
**Dashboard properties:**
| Property | Type | Required | Description |
| ------ | ------ | ------ | ------ |
| `dashboard` | string | yes | Heading for the dashboard. Only one dashboard should be defined per file. |
| `panel_groups` | array | yes | The panel groups which should be on the dashboard. |
**Panel group (`panel_groups`) properties:**
| Property | Type | Required | Description |
| ------ | ------ | ------ | ------ |
| `group` | string | required | Heading for the panel group. |
| `priority` | number | optional, defaults to order in file | Order to appear on the dashboard. Higher number means higher priority, which will be higher on the page. Numbers do not need to be consecutive. |
| `panels` | array | required | The panels which should be in the panel group. |
| `weight` | number | no, defaults to order in file | Order to appear within the grouping. Lower number means higher priority, which will be higher on the page. Numbers do not need to be consecutive. |
| `metrics` | array | yes | The metrics which should be displayed in the panel. Any number of metrics can be displayed when `type` is `area-chart` or `line-chart`, whereas only 3 can be displayed when `type` is `anomaly-chart`. |
| `id` | string | no | Used for associating dashboard metrics with database records. Must be unique across dashboard configuration files. Required for [alerting](#setting-up-alerts-for-prometheus-metrics-ultimate) (support not yet enabled, see [relevant issue](https://gitlab.com/gitlab-org/gitlab-foss/issues/60319)). |
| `label` | string | no, but highly encouraged | Defines the legend-label for the query. Should be unique within the panel's metrics. Can contain time series labels as interpolated variables. |
| `query` | string | yes if `query_range` is not defined | Defines the Prometheus query to be used to populate the chart/panel. If defined, the `query` endpoint of the [Prometheus API](https://prometheus.io/docs/prometheus/latest/querying/api/) will be utilized. |
| `query_range` | string | yes if `query` is not defined | Defines the Prometheus query to be used to populate the chart/panel. If defined, the `query_range` endpoint of the [Prometheus API](https://prometheus.io/docs/prometheus/latest/querying/api/) will be utilized. |
Dynamic labels are useful when multiple time series are returned from a Prometheus query.
When a static label is used and a query returns multiple time series, then all the legend items will be labeled the same, which makes identifying each time series difficult:
For labels to be more explicit, using variables that reflect time series labels is a good practice. The variables will be replaced by the values of the time series labels when the legend is rendered:
The resulting rendered legend will look like this:
There is also a shorthand value for dynamic dashboard labels that make use of only one time series label:
```yaml
metrics:
- id: my_metric_id
query_range: 'http_requests_total'
label: "Method"
unit: "count"
```
This works by lowercasing the value of `label` and, if there are more words separated by spaces, replacing those spaces with an underscore (`_`). The transformed value is then checked against the labels of the time series returned by the Prometheus query. If a time series label is found that is equal to the transformed value, then the label value will be used and rendered in the legend like this:
| query_range | string | required | For area panel types, you must use a [range query](https://prometheus.io/docs/prometheus/latest/querying/api/#range-queries) |
Starting in [version 12.8](https://gitlab.com/gitlab-org/gitlab/issues/202696), the y-axis values will automatically scale according to the data. Previously, it always started from 0.
To add an anomaly chart panel type to a dashboard, add add a panel with *exactly* 3 metrics.
The first metric represents the current state, and the second and third metrics represent the upper and lower limit respectively:
```yaml
dashboard: 'Dashboard Title'
panel_groups:
- group: 'Group Title'
panels:
- type: anomaly-chart
title: "Chart Title"
y_label: "Y-Axis"
metrics:
- id: anomaly_requests_normal
query_range: 'http_requests_total'
label: "# of Requests"
unit: "count"
metrics:
- id: anomaly_requests_upper_limit
query_range: 10000
label: "Max # of requests"
unit: "count"
metrics:
- id: anomaly_requests_lower_limit
query_range: 2000
label: "Min # of requests"
unit: "count"
```
Note the following properties:
| Property | Type | Required | Description |
| ------ | ------ | ------ | ------ |
| type | string | required | Must be `anomaly-chart` for anomaly panel types |
| query_range | yes | required | For anomaly panel types, you must use a [range query](https://prometheus.io/docs/prometheus/latest/querying/api/#range-queries) in every metric. |
To add a column panel type to a dashboard, look at the following sample dashboard file:
```yaml
dashboard: 'Dashboard Title'
panel_groups:
- group: 'Group title'
panels:
- title: "Column"
type: "column"
metrics:
- id: 1024_memory
query: 'avg(sum(container_memory_usage_bytes{container_name!="POD",pod_name=~"^%{ci_environment_slug}-([^c].*|c([^a]|a([^n]|n([^a]|a([^r]|r[^y])))).*|)-(.*)",namespace="%{kube_namespace}"}) by (job)) without (job) / count(avg(container_memory_usage_bytes{container_name!="POD",pod_name=~"^%{ci_environment_slug}-([^c].*|c([^a]|a([^n]|n([^a]|a([^r]|r[^y])))).*|)-(.*)",namespace="%{kube_namespace}"}) without (job)) /1024/1024'
unit: MB
label: "Memory Usage"
```
Note the following properties:
| Property | Type | Required | Description |
| ------ | ------ | ------ | ------ |
| type | string | yes | Type of panel to be rendered. For column panel types, set to `column` |
| query_range | yes | yes | For column panel types, you must use a [range query](https://prometheus.io/docs/prometheus/latest/querying/api/#range-queries) |
| `type` | string | yes | Type of panel to be rendered. For stacked column panel types, set to `stacked-column` |
| `query_range` | yes | yes | For stacked column panel types, you must use a [range query](https://prometheus.io/docs/prometheus/latest/querying/api/#range-queries) |
| type | string | yes | Type of panel to be rendered. For single stat panel types, set to `single-stat` |
| query | string | yes | For single stat panel types, you must use an [instant query](https://prometheus.io/docs/prometheus/latest/querying/api/#instant-queries) |
> [Introduced](https://gitlab.com/gitlab-org/gitlab/issues/201946) in GitLab 12.8.
Query results sometimes need to be represented as a percentage value out of 100. You can use the `max_value` property at the root of the panel definition:
For example, if you have a query value of `53.6`, adding `%` as the unit results in a single stat value of `53.6%`, but if the maximum expected value of the query is `120`, the value would be `44.6%`. Adding the `max_value` causes the correct percentage value to display.
> [Introduced](https://gitlab.com/gitlab-org/gitlab/issues/30581) in GitLab 12.5.
To add a heatmap panel type to a dashboard, look at the following sample dashboard file:
```yaml
dashboard: 'Dashboard Title'
panel_groups:
- group: 'Group Title'
panels:
- title: "Heatmap"
type: "heatmap"
metrics:
- id: 10
query: 'sum(rate(nginx_upstream_responses_total{upstream=~"%{kube_namespace}-%{ci_environment_slug}-.*"}[60m])) by (status_code)'
unit: req/sec
label: "Status code"
```
Note the following properties:
| Property | Type | Required | Description |
| ------ | ------ | ------ | ------ |
| type | string | yes | Type of panel to be rendered. For heatmap panel types, set to `heatmap` |
| query_range | yes | yes | For area panel types, you must use a [range query](https://prometheus.io/docs/prometheus/latest/querying/api/#range-queries) |
### View and edit the source file of a custom dashboard
> [Introduced](https://gitlab.com/gitlab-org/gitlab/issues/34779) in GitLab 12.5.
When viewing a custom dashboard of a project, you can view the original
`.yml` file by clicking on **Edit dashboard** button.
From each of the panels in the dashboard, you can access the context menu by clicking the **{ellipsis_v}** **More actions** dropdown box above the upper right corner of the panel to take actions related to the chart's data.
> [Introduced](https://gitlab.com/gitlab-org/gitlab/-/merge_requests/6590) in [GitLab Ultimate](https://about.gitlab.com/pricing/) 11.2 for [custom metrics](#adding-additional-metrics-premium), and 11.3 for [library metrics](prometheus_library/metrics.md).
For managed Prometheus instances using auto configuration, alerts for metrics [can be configured](#adding-additional-metrics-premium) directly in the performance dashboard.
For manually configured Prometheus servers, a notify endpoint is provided to use with Prometheus webhooks. If you have manual configuration enabled, an **Alerts** section is added to **Settings > Integrations > Prometheus**. This contains the *URL* and *Authorization Key*. The **Reset Key** button will invalidate the key and generate a new one.
To send GitLab alert notifications, copy the *URL* and *Authorization Key* into the [`webhook_configs`](https://prometheus.io/docs/alerting/configuration/#webhook_config) section of your Prometheus Alertmanager configuration:
In order for GitLab to associate your alerts with an [environment](../../../ci/environments.md), you need to configure a `gitlab_environment_name` label on the alerts you set up in Prometheus. The value of this should match the name of your Environment in GitLab.
>- [Introduced](https://gitlab.com/gitlab-org/gitlab/issues/4925) in [GitLab Ultimate](https://about.gitlab.com/pricing/) 11.11.
>- [From GitLab Ultimate 12.5](https://gitlab.com/gitlab-org/gitlab/issues/13401), when GitLab receives a recovery alert, it will automatically close the associated issue.
Once enabled, an issue will be opened automatically when an alert is triggered which contains values extracted from [alert's payload](https://prometheus.io/docs/alerting/configuration/#webhook_config
):
- Issue author: `GitLab Alert Bot`
- Issue title: Extract from `annotations/title`, `annotations/summary` or `labels/alertname`
- Alert `Summary`: A list of properties
-`starts_at`: Alert start time via `startsAt`
-`full_query`: Alert query extracted from `generatorURL`
- Optional list of attached annotations extracted from `annotations/*`
When GitLab receives a **Recovery Alert**, it will automatically close the associated issue. This action will be recorded as a system message on the issue indicated that it was closed automatically by the GitLab Alert bot.
To further customize the issue, you can add labels, mentions, or any other supported [quick action](../quick_actions.md) in the selected issue template, which will apply to all incidents. To limit quick actions or other information to only specific types of alerts, use the `annotations/gitlab_incident_markdown` field.
Since [version 12.2](https://gitlab.com/gitlab-org/gitlab-foss/issues/63373), GitLab will tag each incident issue with the `incident` label automatically. If the label does not yet exist, it will be created automatically as well.
If the metric exceeds the threshold of the alert for over 5 minutes, an email will be sent to all [Maintainers and Owners](../../permissions.md#project-members-permissions) of the project.
It is possible to display metrics charts within [GitLab Flavored Markdown](../../markdown.md#gitlab-flavored-markdown-gfm) fields such as issue or merge request descriptions. The maximum number of embedded charts allowed in a GitLab Flavored Markdown field is 100.
It is also possible to embed either the default dashboard metrics or individual metrics in issue templates. For charts to render side-by-side, links to the entire metrics dashboard or individual metrics should be separated by either a comma or a space.
### Embedding Cluster Health Charts **(ULTIMATE)**
> [Introduced](<https://gitlab.com/gitlab-org/gitlab/issues/40997>) in [GitLab Ultimate](https://about.gitlab.com/pricing/) 12.9.
[Cluster Health Metrics](../clusters/index.md#monitoring-your-kubernetes-cluster-ultimate) can also be embedded in [GitLab-flavored Markdown](../../markdown.md).
To embed a metric chart, include a link to that chart in the form `https://<root_url>/<project>/-/cluster/<cluster_id>?<query_params>` anywhere that GitLab-flavored Markdown is supported. To generate and copy a link to the chart, follow the instructions in the [Cluster Health Metric documentation](../clusters/index.md#monitoring-your-kubernetes-cluster-ultimate).
The following requirements must be met for the metric to unfurl:
- The `<cluster_id>` must correspond to a real cluster.
- Prometheus must be monitoring the cluster.
- The user must be allowed access to the project cluster metrics.
- The dashboards must be reporting data on the [Cluster Health Page](../clusters/index.md#monitoring-your-kubernetes-cluster-ultimate)
If the above requirements are met, then the metric will unfurl as seen below.
It is possible to embed live [Grafana](https://docs.gitlab.com/omnibus/settings/grafana.html) charts in issues, as a [direct linked rendered image](https://grafana.com/docs/grafana/latest/reference/share_panel/#direct-link-rendered-image).
Copy the link and add an image tag as [inline HTML](../../markdown.md#inline-html) in your Markdown. You may tweak the query parameters as required. For instance, removing the `&from=` and `&to=` parameters will give you a live chart. Here is example markup for a live chart from GitLab's public dashboard:
Each project can support integration with one Grafana instance. This configuration allows a user to copy a link to a panel in Grafana, then paste it into a GitLab Markdown field. The chart will be rendered in the GitLab chart format.
1. In Grafana, click on a panel's title, then click **Share** to open the panel's sharing dialog to the **Link** tab. If you click the _dashboard's_ share panel instead, GitLab will attempt to embed the first supported panel on the dashboard (if available).
1. If your Prometheus queries use Grafana's custom template variables, ensure that "Template variables" option is toggled to **On**. Of Grafana global template variables, only `$__interval`, `$__from`, and `$__to` are currently supported. Toggle **On** the "Current time range" option to specify the time range of the chart. Otherwise, the default range will be the last 8 hours.
- You may need to re-add the GitLab predefined common metrics. This can be done by running the [import common metrics rake task](../../../administration/raketasks/maintenance.md#import-common-metrics).
