You add variables via Dashboard cogs menu > Templating. This opens up a list of variables and a `New` button to create a new variable.
### Basic variable options
Option | Description
------- | --------
*Name* | The name of the variable, this is the name you use when you refer to your variable in your metric queries. Must be unique and contain no white-spaces.
*Label* | The name of the dropdown for this variable.
*Query* | This variable type allows you to write a data source query that usually returns a list of metric names, tag values or keys. For example, a query that returns a list of server names, sensor ids or data centers.
*Interval* | This variable can represent time spans. Instead of hard-coding a group by time or date histogram interval, use a variable of this type.
*Datasource* | This type allows you to quickly change the data source for an entire Dashboard. Useful if you have multiple instances of a data source in for example different environments.
*Custom* | Define the variable options manually using a comma separated list.
*Constant* | Define a hidden constant. Useful for metric path prefixes for dashboards you want to share. During dashboard export, constant variables will be made into an import option.
*Ad hoc filters* | Very special kind of variable that only works with some data sources, InfluxDB & Elasticsearch currently. It allows you to add key/value filters that will automatically be added to all metric queries that use the specified data source.
*Refresh* | Controls when to update the variable option list (values in the dropdown). **On Dashboard Load** will slow down dashboard load as the variable query needs to be completed before dashboard can be initialized. Set this only to **On Time Range Change** if your variable options query contains a time range filter or is dependent on dashboard time range.
*Include All option* | Add a special `All` option whose value includes all options.
*Custom all value* | By default the `All` value will include all options in combined expression. This can become very long and can have performance problems. Many times it can be better to specify a custom all value, like a wildcard regex. To make it possible to have custom regex, globs or lucene syntax in the **Custom all value** option it is never escaped so you will have to think avbout what is a valid value for your data source.
**Elasticsearch** uses lucene query syntax, so the same variable would, in this case, be formatted as `("host1" OR "host2" OR "host3")`. In this case every value
*Tags query* | Data source query that should return a list of tags
*Tag values query* | Data source query that should return a list of values for a specified tag key. Use `$tag` in the query to refer the currently selected tag.
Use the `Interval` type to create a variable that represents a time span (eg. `1m`,`1h`, `1d`). There is also a special `auto` option that will change depending on the current time range. You can specify how many times the current time range should be divided to calculate the current `auto` timespan.
This variable type is useful as a parameter to group by time (for InfluxDB), Date histogram interval (for Elasticsearch) or as a *summarize* function parameter (for Graphite).
Example using the template variable `myinterval` of type `Interval` in a graphite function:
```
summarize($myinterval, sum, false)
```
## Global Built-in Variables
Grafana has global built-in variables that can be used in expressions in the query editor.
### The $__interval Variable
This $__interval variable is similar to the `auto` interval variable that is described above. It can be used as a parameter to group by time (for InfluxDB), Date histogram interval (for Elasticsearch) or as a *summarize* function parameter (for Graphite).
Grafana automatically calculates an interval that can be used to group by time in queries. When there are more data points than can be shown on a graph then queries can be made more efficient by grouping by a larger interval. It is more efficient to group by 1 day than by 10s when looking at 3 months of data and the graph will look the same and the query will be faster. The `$__interval` is calculated using the time range and the width of the graph (the number of pixels).
For example, when the time range is 1 hour and the graph is full screen, then the interval might be calculated to `2m` - points are grouped in 2 minute intervals. If the time range is 6 months and the graph is full screen, then the interval might be `1d` (1 day) - points are grouped by day.
In the InfluxDB data source, the legacy variable `$interval` is the same variable. `$__interval` should be used instead.
The InfluxDB and Elasticsearch data sources have `Group by time interval` fields that are used to hard code the interval or to set the minimum limit for the `$__interval` variable (by using the `>` syntax -> `>10m`).
This variable is the `$__interval` variable in milliseconds (and not a time interval formatted string). For example, if the `$__interval` is `20m` then the `$__interval_ms` is `1200000`.
### The $timeFilter or $__timeFilter Variable
The `$timeFilter` variable returns the currently selected time range as an expression. For example, the time range interval `Last 7 days` expression is `time > now() - 7d`.
This is used in the WHERE clause for the InfluxDB data source. Grafana adds it automatically to InfluxDB queries when in Query Editor Mode. It has to be added manually in Text Editor Mode: `WHERE $timeFilter`.
The `$__timeFilter` is used in the MySQL data source.
This variable is only available in the Singlestat panel and can be used in the prefix or suffix fields on the Options tab. The variable will be replaced with the series name or alias.
Only make changes to the first panel (the original template). To have the changes take effect on all panels you need to trigger a dynamic dashboard re-build.
You can do this by either changing the variable value (that is the basis for the repeat) or reload the dashboard.
