# filter Metrics Operator

The functionality provided by the `filter`

operator has been incorporated into the `where`

operator. We recommend the use of `where`

over `filter`

, because `filter`

will be deprecated in the future. For more information, see `where`

Metrics Operator

You can use the `filter`

operator to limit the results returned by a metric query. There are several ways you can restrict results. You can apply an aggregation function, such as `avg`

, to a time series. You can also filter based on how many times the value of individual data points meet a value condition over a particular duration.

## Syntax

There are two supported syntaxes for the `filter`

operator.

`filter [REDUCER BOOLEAN EXPRESSION]`

or

`filter _value [VALUE BOOLEAN EXPRESSION] [all | atleast n] [first | any | last] [duration]`

### Syntax 1

The first variant filters based on a function (usually an aggregation function) applied to the time series.

`filter [REDUCER BOOLEAN EXPRESSION]`

Where:

`[REDUCER BOOLEAN EXPRESSION]`

is an expression that takes all the values of a given time series, uses a function to reduce them to a single value, and evaluates that value.

The supported functions are:

`avg`

. Returns the average of the time series.`min`

. Returns the minimum value in the time series.`max`

. Returns the maximum value in the time series.`sum`

. Returns the sum of the values in the time series.`count`

. Returns the count of data points in the time series.`pct(n)`

. Returns the nth percentile of the values in the time series.`latest`

. Returns the last data point in the time series.

#### Syntax 1 examples

**Example 1**

Return the time series in which the average value of the CPU_User metric is greater than 95:

`metric=CPU_User | filter avg > 95``

**Example 2**

Return the time series in which the latest value of the CPU_User metric is greater than 50:

`metric=CPU_User | filter latest > 50`

### Syntax 2

The second variant filters based on how many times the values of individual data points of a time series meet a value condition over a particular duration.

`filter _value [VALUE BOOLEAN EXPRESSION] [all | atleast n] [first | any | last] [duration]`

Where:

`[VALUE BOOLEAN EXPRESSION]`

is a value expression that operates on individual data points of a time series. For example,`> 3`

- Use
`all`

to specify that all data points within the duration must meet the value condition, or`atleast n`

, where`n`

is a count, to specify how many data points must meet the value condition. - Use
`first`

,`any`

, or`last`

to specify what part of the time range that duration applies to: the start of the time range, any part of the time range, or the end of the time range. - Use
`duration`

to specify the length of time to consider in the query in minutes (m), hours (h), or days (d). For example,`5m`

,`6h`

, or`1d`

.

#### Syntax 2 examples

**Example 1**

Return only the time series in which all data points during the last 5 minutes of the query time range have a value greater than 3.

There must be a least one data point in the last 5 minutes of the time range for this to be valid.

`metric=CPU_User | filter _value > 3 all last 5m`

**Example 2**

Return only the time series that have at least 1 data point greater than 3 for the last 5 minutes of the query time range.

`metric=CPU_User | filter _value > 3 atleast 1 last 5m`

**Example 3**

Return only the time series that have only values greater than 3 for any consecutive 5 minutes of the time range.

`metric=CPU_User | filter _value > 3 all any 5m`

**Example 4**

Return only the time series that have only values greater than 3 for the first 5 minutes of the query time range.

There must be a least one data point in the first 5 minutes of the time range for this to be valid.

`metric=CPU_User | filter _value > 3 all first 5m`