Splunk Search
Highlighted

Does stats distinct_count have a limit of distinct values it will count?

SplunkTrust
SplunkTrust

I was working with a search similar to:

my_nifty_search_terms | stats distinct_count(field) by date_hour

and noticed that at least one of my counts stopped conveniently at 100,000. Is this perhaps a hard coded limit somewhere?

Tags (1)
Highlighted

Re: Does stats distinct_count have a limit of distinct values it will count?

Splunk Employee
Splunk Employee

It is. The limit has been removed in either 4.1.6 or will be removed in 4.2. (Update: limit removed as of 4.1.6.) In the meantime, you can instead do:

my_nifty_search_terms | stats count by field,date_hour | stats count by date_hour

This will not be subject to the limit even in earlier (4.x) versions.


This limit does not exist as of 4.1.6, so you can use distinct_count() (or dc()) even if the result would be over 100,000.

In older versions (4.1.5 and down), you can use:

... | stats distinct_count(f)

can be replaced with

... | stats count by f | stats count

and in general

... | stats dc(f) as g by x,y,z,a,...

can be replaced with

... | stats count as g by f,x,y,z,a,... | stats count as g by x,y,z,a,...

View solution in original post