Skip to main content

Group or Aggregate Operators

Aggregating functions evaluate messages and place them into groups. The group operator is used in conjunction with group-by functions.

Only search results that have been aggregated using a group or aggregate operator can be placed on a dashboard panel. See Chart Search Results for information about charting.

When using any grouping function, the word by is sufficient for representing the group operator. The typical construction when using group-by functions is:

grouping_function by <fieldname>

Aggregating (group-by) functions include:

important

The withtime, most_recent, and least_recent operators are not considered standalone operators; they are designed to only be used as an alternative to the first and last operators in Live Dashboards or any continuous query where first and last are not supported.

By default, the ordering is not defined inside of groups created using a group-by expression. To order your results, use the sort operator.

Syntax​

  • ... | group_by_function <field_to_operate_on> group by <field_to_group_by>[, <field2>, ...]

You can use by instead of group by so count group by user is equivalent to count by user.

Rules​

  • Can not be used with the LogReduce operator.
  • When parsing and naming (aliasing) fields, avoid using the names of grouping functions or other operators as field names.
  • When using count, or any grouping function, remember to include the underscore before the field name (sort by _count).
  • Multiple aggregation functions can be on the same line but you can't include another function, such as a math function, on the same line of a query.

For example, you can't use:

... | avg(x + y) as average, sum(x+y) as total

You would need to do that in two separate steps, such as:

... | x + y as z | avg(z) as average, sum(z) as total

In another example, you can't use:

avg(abs_latency)/1000/60 as avg_latency_min

Instead, you'd need to use two separate lines:

avg(abs_latency_ms) as avg_latency_ms 
| avg_latency_ms / 1000 / 60 as avg_latency_min

Examples​

Sort by _count and limit to 10 results​

* | parse "GET * " as url  
| count by url  
| sort by _count  
| limit 10

Count by user​

status AND down  
| parse regex "user=(\<use\>.*?)" 
| parse regex "host=(\<msg_hos\>.*?)" 
| count by user

Count by the Source IP address​

_sourceCategory=apache  
| parse "* " as src_ip 
| parse "GET *" as url 
| count by src_ip 
| sort by _count

Group by multiple fields​

| count(field1), avg(field2) group by field1, _timeslice

Use multiple aggregate operators​

| max(amount) as amount_max, count(datetime) as datetime_count, sum(_size) as messages_size_sum, last(query) as last_query

All Sumo Logic system-generated fields begin with an underscore ("_"). Group-by functions always create a Sumo Logic field named with a combination of an underscore (_) and the function name. Using the function count inserts a field into the pipeline called _count. The function count_distinct inserts a field into the pipeline called _count_distinct.