Skip to content

Variable Queries

View variable queries are used to define the value range and data source of variables. They support configuring whether the value range follows the dashboard time widget, meeting different data filtering needs.

Value Range Control

The selectable values obtained from the query can be chosen to be affected by the dashboard time range or not:

Mode Description Use Cases
Follow Dashboard Linkage Filter data based on the dashboard time widget View only recently active hosts to avoid historical garbage data
Do Not Follow Dashboard Linkage List all historical data Need full enumeration, such as all service names (services usually do not change frequently)

Main Query Methods

View variables currently support five types of query methods. You can choose as needed:

Query Method Use Cases Key Features
DQL Query Querying non-metric data such as LOG, objects, APM, etc. The only method supporting all data types, supports cascade queries
PromQL Query Querying only metric data and familiar with Prometheus Syntax compatible with Prometheus ecosystem
Data Type Query Quick point-and-click configuration without writing queries Visual selection, zero code
External Data Source Connecting to external databases like MySQL Directly use SQL queries
Custom Only a few fixed options Directly fill in, no database query

DQL Query

Supports all data types including Metrics, LOG, Infrastructure, Objects, APM, RUM, etc. It is the most comprehensive.

Click Learn more about DQL definitions and syntax.

Basic Syntax Structure

Data Type::Data Source:(distinct('Attribute Field Name')){Filter Conditions}

Example:

# Get hostnames of all Docker containers
O::docker_containers:(distinct(`host`))

Specifying Time Range

DQL supports directly specifying the time range in the statement, which takes precedence over the dashboard time widget:

Syntax Meaning
[10m] Last 10 minutes
[1h] Last 1 hour
[1d] Last 1 day
Not written Use dashboard time widget

Example:

# Only show hosts with data in the last 10 minutes
O::docker_containers:(distinct(`host`)) [10m]

Show Function Query

Suitable for querying metric-related metadata information. Does not support time range.

Syntax Structure
Syntax Description
show_measurement() Returns all measurements in the current workspace
show_tag_key(from=["cpu"]) Returns tags under the measurement cpu in the current workspace
show_tag_key(from=["cpu"],keyin=["host"]) Queries the list of host tag values for the measurement cpu in the current workspace

Click Learn more about DQL function descriptions.

Example:

# Query all measurements
SHOW METRIC_SET()

# Query all tags for a specified measurement
SHOW LABELS FROM M::cpu

PromQL Query

PromQL queries support handwritten query statements. By default, they use the time selected by the dashboard time widget to query and return data.

For information about PromQL queries, refer to PromQL Quick Start.

Data Type Query

If you need quick configuration and don't want to write query statements, you can directly select the data type and choose filter conditions on the page.

Supported data types:

  • Metrics
  • Basic Objects (hosts, containers, processes, etc.)
  • Resource Catalog
  • LOG
  • APM
  • RUM

In this mode, you cannot add time ranges or configure cascade queries.

External Data Source

If your target query selectable values are stored in an external database (e.g., MySQL, PostgreSQL), you can choose this query method to directly connect to the existing data system.

  1. Select the data source;
  2. Enter the query statement;
  3. Select the sorting method.

For more details, refer to External Data Sources.

Custom Query

If your target selectable values are fixed and limited in number, and there's no need to query the database, you can directly define a set of values for the view variable. Separate each option with an English comma ,, e.g., prod,test,dev,staging.

This query method does not support adding time ranges or configuring cascade queries.

Cascade Query

Used when multiple variables have a hierarchical relationship. For example:

  • After selecting "service", the "environment" dropdown only shows environments for that service (not all environments).
  • After selecting "environment", the "version" dropdown only shows versions for that environment.

Configuration Conditions

  1. Only DQL queries support cascade configuration;
  2. In the child variable's query statement, reference the parent variable's current value using #{parent variable name}.

Configuration Example

The following uses the cascade filtering of view variables in the Web Application Overview of RUM as an example, illustrating the configuration for linkage queries based on service, environment, and version:

Three variable query statements need to be configured in the dashboard. The second and third query statements use #{variable name} configured in the first query:

  • Variable 1 (Service): Query all application lists:
T::re(`.*`):(distinct(service))
  • Variable 2 (Environment): Query all application lists, and the environment list selected in Variable 1:
T::re(`.*`):(distinct(env)) { service = '#{service}'}
  • Variable 3 (Version): Query all application lists, and the version list selected in Variable 2:
T::re(`.*`):(distinct(version)) {service = '#{service}'}

Query Scenario Examples

Associating Charts

  1. In the chart query, when filtering fields, select the variable for the value;

  2. Switch the variable value in the scenario view, and the chart will be filtered and displayed based on the variable tag.

Chart Reverse Addition

Prerequisite

The corresponding DQL filter contains a by grouping condition.

  1. After clicking on a specific time series or data point in the chart;
  2. Select Apply to View Variable;
  3. The system automatically fills in the query.