Joins two LogScale searches. When joining two searches, you need to define the keys/fields that are used to match up results. This is done using the field=name or field=[name,name,...] parameter. If you want to join on a single field name, you can use the syntax

fieldName =~ join(...)

to specify the field.

If the subquery has a different field that you want to match against, then use the parameter key=[name1,name2,...] to designate the names of keys inside the subquery. The value of keys defaults to the value of field.

join() is a filter function which in the default mode=inner lets the events through that match on the join keys. If you specify mode=left then events that do not match the join key(s) will also be let through.

If you specify include=[field, field, ...] then those fields are extracted from the result of the subquery, and added to matching events. For events in the subquery that do not have one or more of the named include fields, the output will be the empty string.

Using the parameter max=N (which defaults to max=1) you can specify how many rows/events are picked up in the subquery. If a subquery has multiple events with the same join key, then up to max rows are emitted.

You can use the parameters start and end to specify an alternative time interval for the query. The parameter view can be used to direct the subquery to run in a different repository or view, and the live=true|false parameter can be used to control if the subquery runs as a live query. The defaults for all these parameters are inherited from the main query containing the join(...) usage.

The join() function also has a concept of a maximum size of the resultset of the inner query specified with the limit=100000 parameter.


The join() function does two passes over the data and can therefore not run truly live.

When used in a live query, the query will be run in a repeated mode instead, in which the server chooses the repetition interval based on the resources used by the function.

This can impact the liveness of the query, in that long-running repeated queries can be throttled, and thus be less live than expected.

endstringfalseEnd of main queryEnd of time interval of subquery (milliseconds since UTC or 2d, 24h, 10sec, etc).
field[string]true Specifies which field in the event (log line) must match the given column value.
include[string]falsenoneSpecifies columns to include from the subquery.
key[string]false Specifies which fields of the subquery to join on. Defaults to the value of the field parameter.
limitnumberfalse100000Specifies the maximum number of rows in the subquery.
livebooleanfalseSame as main queryControl if the subquery runs as live or static query.
maxnumberfalse1Maximum number of events found in subquery if several share join key.
modestringfalseinnerSpecifies the mode (inner or left) of the join.
  Valid Valuesinner 
queryFunctionfalse The subquery to execute producing the values to join with. [a]
repostringfalseRepo of main querySpecify which view/repo in which to perform the subquery.
startstringfalseStart of main queryStart of time interval of subquery (milliseconds since UTC or 2d, 24h, 10sec, etc).
viewstringfalseView of main querySpecify which view/repo in which to perform the subquery.

[a] If an argument name is not given, query is the default argument.

Find some examples at Joins section.