> ## Documentation Index
> Fetch the complete documentation index at: https://docs.chronosphere.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Metrics Explorer
export const AddToDashboardIcon = props => {
return ;
};
**Metrics Explorer** in Chronosphere Observability Platform helps you craft, investigate,
and troubleshoot queries of your collected metrics, and also visualizes the results.
Use Metrics Explorer to help create and polish the results of queries before
you add them to a [dashboard](/observe/dashboards) or [monitor](/investigate/alerts/monitors).
To access Metrics Explorer, in the navigation menu, click
** Explorers > Metrics Explorer**.
## Enter and run a query
Metrics Explorer defaults to querying Prometheus metrics.
Like the rest of Observability Platform, Metrics Explorer supports the
[Prometheus Query Language (PromQL)](/investigate/querying/promql) for writing
queries. For full details about PromQL, read the
[Prometheus documentation](https://prometheus.io/docs/prometheus/latest/querying/basics/).
The query field supports autocompletion for metric names and functions.
To enter and run a query:
1. In the **query field** begin typing to reveal a list of matching results, and
either click a result or use the arrow keys and `Enter`
(`Return` on macOS) to add it to the query.
2. Optional: Click **Builder** to open the
[Query Builder](/investigate/querying/metrics/query-builder) which you can use to
construct, optimize, and debug queries before saving and using them, and for
sharing queries with your team.
3. To add an additional query to execute, click ** Add query**.
Each query adds a series to Metrics Explorer's **Graph** and **Table**.
4. To execute all of the entered queries, click ** Run**,
or press `Alt+Enter` (`Command+Return` on macOS) on your
keyboard.
To delete a query, click its ** Delete query**
button.
### Mark a query as a dynamic threshold
This feature isn't available to all Chronosphere Observability Platform users and
might not be visible in your app. For information about enabling this feature in your
environment, contact [Chronosphere Support](/support).
Mark an individual query as a *dynamic threshold* to render its returned series
as a dashed reference line overlaid on the graph rather than as a regular data
series. Use dynamic thresholds to display reference values alongside your primary
data, such as an SLO target or a historical baseline.
To mark a query as a dynamic threshold:
1. In the **Edit panel** interface, click the **Query** tab.
2. Next to the query you want to use as a threshold, click the
**Render query as threshold** button. The query is marked as a threshold
and its series render differently from regular data series.
3. Optional: Click the colored circle icon that appears to open a color picker
and choose a custom threshold color.
4. Click **Apply**, and then click **Save** to save the dashboard.
To remove the threshold flag from a query, click the colored circle icon next to
the query, and then click the ** Remove threshold from query**
icon in the color picker.
### Automatically execute queries based on an interval
To refresh data automatically, see
[Refresh a time range](/navigate/time-ranges#refresh-a-time-range).
### Generate queries from natural language prompts
Observability Platform provides
[generative artificial intelligence](/overview/generative-ai) (AI) tools to generate
queries from natural language prompts. Using natural language, you write what you
want Observability Platform to query, and Observability Platform generates the query
that returns relevant data.
You can generate and refine PromQL queries from natural language prompts in
Metrics Explorer. Text entry fields that support natural language prompts include
an **Edit with AI** button. When you generate a query,
[semantic search](/investigate/querying/natural-language#semantic-metric-search)
identifies the most relevant metrics for your prompt.
For more information on how to generate and refine queries with AI, see
[Generate queries using natural language](/investigate/querying/natural-language).
Generative AI features can produce incorrect results, hallucinate data, and deliver
inaccurate analysis. Use generative AI features with care, and independently verify
all information produced by generative AI tools before applying it.
Certain prompts, data, or other inputs might produce irrelevant content. Don't rely on
generative AI features or responses for any uses that exceed their designed scope.
## Search metrics and labels
Click **Browse metrics** next to the **query field** to open the **Explore metrics**
panel, which displays a list of available metrics and a search field.
From there, you can ** Insert** or
** Copy** the names of metrics, labels, or
values that you browse, filter, and select from this list.
Begin typing in the search field to filter the metrics by keyword. The view lists
the first 100 matching metrics.
Each row of the results represents a metric and includes the following functions:
* ** Explore labels**: Opens a list of that metric's
labels, with filters you can engage to select specific values associated with
that label. For details, see [Explore labels](#explore-labels).
* ** Insert**: Inserts the metric name
into the **query field** at the cursor's position.
* ** Copy to clipboard**: Copies the metric
name into your clipboard.
## Explore labels
When you click ** Explore labels** for a metric in the
**Explore metrics** panel, Metrics Explorer lists all labels associated with that
metric, and the values associated with each label. It also lists the **selector**,
which is the query text representing currently selected metric or label value.
### Filter label values by match
Each label name has a **Filter** button. To filter label values by conditionally
matching text:
1. Click **Filter** on the label you want to filter values for.
2. Optional: To change the operator, click the operator dropdown, which defaults to
**`=` Equal to**.
* **`=` Equal to**: Matches only if the value exactly matches the entered text.
* **`!=` Not equal to**: Matches only if the value doesn't exactly match the
entered text.
* **`=~` Regular expression**: Matches only if the value matches the entered
regular expression.
* **`!~` Negative regular expression**: Matches only if the value doesn't match
the entered regular expression.
3. Enter text or a regular expression, depending on the selected operator. For text
operators, Metrics Explorer provides autocompletion suggestions as you type from
the list of known values.
4. Click **Apply** to apply the filter, or ** Close** to
cancel and clear the filter.
The entered filter replaces the **Filter** button, and the list of label values
is now filtered accordingly. This also updates the **selector** with the filter.
To remove the applied filter, click the ** Remove filter**
button, either where it's adjacent to the label on the list or at the **Filters**
list in the header.
### Filter label values by value
In addition to text and regular expression filters, you can filter by a specific
label value. Each value listed with each label is a link, and clicking it applies
that label name as if it were an **`=` Equal to** filter.
### Use the selector in your query
As with the metrics list, you can click ** Insert**
and ** Copy** to respectively insert the
selector into the query or copy it to your clipboard.
## Modify query visualizations
Metrics Explorer's main navigation bar includes controls similar to
[dashboards](/observe/dashboards) for comparing past values, and for defining and
changing time ranges used in Metrics Explorer's visualizations.
To use Metrics Explorer's **Compare** tool to compare current values to past values,
and to define and change time ranges for visualizations using the
** [time range selector](/navigate/time-ranges)**. See
[Compare current data to past data](/navigate/time-ranges#compare-current-data-to-past-data)
for more information.
Use the **Resolution** field change the query resolution. If you have opened this
Metrics Explorer page from a [dashboard panel](/observe/dashboards/panels), the
resolution value carries over from the panel.
Metrics Explorer also provides its own tools and options to modify its
visualizations.
### Visualization settings
Metrics Explorer includes visualizations settings for its charts. Use these settings
to customize the **Display** and the **Null behavior**.
**Display** options change the type of chart.
* **Line** (Default)
* **Area**
* **Bar**
* **Heatmap**
Select an item from the **Null behavior** menu to change how the chart handles null
values:
* **Default**
* **Connected**
* **Null as zero**
Heatmap visualizations don't use null behaviors.
### Hide a query
To hide a query from the Explorer's visualizations, click its
** Hide query** button.
### Split queries into separate graphs
To display each query in a separate graph, enable the **One graph per query** toggle.
Disabling this toggle merges the queries back into a single graph.
### Assign an alias to a query
To assign an alias to a query that appears in the visualization's legend in place
of its query string, click that query's **Alias** button and enter the alias into
the field.
## Modify queries with options
Each query has ** Options** that you can modify
to change the query's naming pattern or step parameter, or to change the query itself.
### Name a query result's series
You can name the visualized result of the query by selecting a method from the
**Series naming** dropdown in a query's options.
1. Click ** Options** for the query you want to
modify.
2. Click the **Series naming** dropdown, which defaults to **Labels**.
3. Select from either **Labels**, to name the series by selecting labels to include,
or **Regex**, to name the series by choosing capture groups to include from a
regular expression.
4. Enter an appropriate pattern.
* For **Labels**, enter a label as `{{ label_name }}` in the **Naming pattern** field
to use the label's value as the series' name.
For example, given a label `env` with values `staging-1`,
`prod-1`, and so on, enter `{{ env }}` as the **Naming pattern** to replace
the series' name in visualizations with the `env` label's corresponding values.
* For **Regex**, enter a regular expression that includes a capture group (`()`)
in the **Regex** field that appears. In the **Naming pattern** field, use `$1`
syntax to substitute the capture group's match into the series name.
For example, given the series `production-123` and `production-456`, enter
`production-(.*)` in the **Regex** field to match both series, then enter
`Prod: $1` as the **Naming pattern** to name those series as `Prod: 123` and
`Prod: 456`.
### Define a query's minimum step period
The PromQL `step` parameter controls how many data points are returned in a range
query.
To define the step parameter in a Metrics Explorer query:
1. Click ** Options** for the query you want to
modify.
2. Enter a time value using in the **Min step** field, using Observability Platform's
[time value syntax](/navigate/time-ranges).
For example, entering a **Min step** of `5s` sets the step value to 5 seconds.
### Define a query's downsampling strategy
You can reduce the resolution of a query's results by selecting a downsampling strategy.
This automatically adds an
[aggregation over time](https://prometheus.io/docs/prometheus/latest/querying/functions/#aggregation_over_time)
function and interval period to the query. Use downsampling to reduce the impact
of high-cardinality queries, such as on visualization performance.
When you apply a downsampling strategy to a query, Observability Platform modifies
the original query to add the corresponding function and interval, and also displays
the *resolved query* without modifying the original query. If you modify the original
query, Observability Platform automatically updates the resolved query.
To define a downsampling strategy:
1. Click ** Options** for the query you want to
modify.
2. Click the **Downsampling strategy** dropdown and select a strategy, which adds
the corresponding function and a time interval to the query:
* **Off**: Reverts to using the original query.
* **Avg**: `avg_over_time`
* **Sum**: `sum_over_time`
* **Min**: `min_over_time`
* **Max**: `max_over_time`
## Add queries to a dashboard
Metrics Explorer supports
[adding queries to a dashboard](/investigate/querying/dashboard-query)
so you can visualize query results.
## Investigate suspicious metrics
Click the three vertical dots icon and then click
[**Metrics DDx**](/investigate/analyze/differential-diagnosis/metrics) to open a query
in the differential diagnosis page.
See [common panel elements](/observe/dashboards/panels#common-panel-elements) for an
explanation of the other available tools.
## View past queries
To view a list of queries that you've recently used in Metrics Explorer, click
** View queries** in the main navigation bar. This opens
the **Queries** panel to the **Recent** tab, which lists recent queries in order
from most to least recent.
When you move the cursor over each query, you can click the
** Dots icon** that appears to **Copy** or **Delete**
the query.
To view [saved queries](#save-queries), click the **Saved** tab in the **Queries**
panel.
To exit the panel, click the panel's ** Close** button.
## Save queries
To save your entered queries, click **Save queries** in the main navigation bar.
This opens the **Queries** panel to the **Saved** tab and proposes saving the queries
by prompting you to enter a name for the queries.
After entering a name, click **Save** to save the queries. Click **Cancel** to remove
the proposed queries from the panel.
To exit the panel, click the panel's ** Close** button.
This discards and proposed but unsaved queries.
## Troubleshooting queries
You can change the following settings for query history from the **Settings** tab:
* The period of time to save query history (default: 1 week)
* The default active tab (default: Query history tab)
* Only show queries for active data source (default: true)
* Clear query history
## Inspector
The **Inspector** helps you understand and troubleshoot queries. Available options
are:
* An overview of **Stats** for the query, including:
* **Total request time**
* **Data processing time**
* **Number of queries**
* **Total number rows**
* The **Query inspector**, which lets you to inspect the raw data.
* The **JSON** tab, to export the query as JSON.
* The **Data** tab, which shows raw data. Click **Download CSV** to export the data
to as a comma-separated values (CSV) file.
## Available metrics for troubleshooting
Additional metrics in your environment track the overall health of alerting and
recording rules that you've configured. The following examples are based on
Prometheus queries and troubleshooting.
Each metric has multiple labels you can use for slicing and monitoring, in the
following format:
* `metric_name`: metric\_description
* `label_name`: label\_description + use
Visit [Prometheus metric naming recommendations](/ingest/metrics-traces/collector/mappings/prometheus/prometheus-recommendations)
for more details about naming metrics.
Observability Platform provides the following metrics for troubleshooting:
* `prometheus_rule_group_last_duration_seconds`: A gauge metric that holds the total
time the group took to complete its last iteration, in seconds.
* `rule_group`: The group that this rule belongs to.
* `prometheus_rule_evaluation_duration_seconds`: A summary metric to track the
average time an individual rule takes to evaluate.
* `prometheus_rule_evaluations_total`: The total number of individual rule
evaluations that occur.
* `rule_group`: The group that this rule belongs to.
* `prometheus_rule_group_iterations_missed_total`: The total number of rule group
evaluations missed due to slow rule group evaluation.
* `rule_group`: The group that this rule belongs to.
* `prometheus_rule_group_iterations_total`: The total number of scheduled rule group
evaluations, whether executed or missed.
* `rule_group`: The group that this rule belongs to.
* `prometheus_rule_eval_failures_total`: The total number of individual rule
evaluation failures.
* `rule_group`: The group that this rule belongs to.
* `type`: Alerting or recording depending on the type of rule.
* `identifier`: The slug for the given rule.
* `status_code`: The status code associated with the given evaluation failure.
### Examples
The following examples explain how to create alerts for particular situations:
* Consistent rule failures
To receive an alert whenever an individual rule consistently fails for five
minutes, [create a monitor](/investigate/alerts/monitors#create-a-monitor)
with the following query:
```text theme={null}
sum by (identifier) (rate(prometheus_rule_eval_failures_total[1m]))
```
with a **Sustain** of 5m.
You can also create an alert for monitoring individual rule failures:
```text theme={null}
sum by (identifier) (rate(prometheus_rule_eval_failures_total{type=""}[1m])
```
* Create alerts by rule type
If you want to alert on certain types of rules, you can do something like:
```text theme={null}
sum by (identifier) (rate(prometheus_rule_eval_failures_total{identifier=~""}[1m]))
```
For example, to create an alert for only a specific category of rules, you can do
something like:
```text theme={null}
sum by (identifier) (rate(prometheus_rule_eval_failures_total{type=""}[1m])
```