When using "tstats count", how to display zero results if there are no counts to display?

[jsh315]

I need to use tstats vs stats for performance reasons. I would like tstats count to show 0 if there are no counts to display.

| tstats count where index="abc" by _time span=1h

This would return a table with _time (every hour) and count, like this:

_time                   count
2017-01-01 12:00        50
2017-01-01 13:00        62
2017-01-01 14:00        14

But if the count is zero, it doesn't return that row. How do I get this to display 0 count for that hour?

Thank you.

[rjthibod · Accepted solution]

How about this

 | tstats count where index="abc" by _time span=1h | makecontinuous span=1h _time | fillnull value=0

[jsh315]

Did not work. Still getting empty rows for where count is zero.

[rjthibod]

Try this instead

  | tstats count where index="abc" by _time span=1h | timechart span=1h max(count) as count | fillnull value=0 count

[jsh315]

That worked. If you have time I would greatly appreciate the explanation of your query. Thank you so much.

[rjthibod]

timechart by default (unless you specify fixedrange=f) creates a row for each time bucket from the beginning of the search period until the end of the search period. So, the timechart creates all the necessary rows, and then fillnull puts a 0 in all empty row.

[jsh315]

Thanks for the response. Your solution works if there is at least one row that returned some count but if all the rows had zero count then I get "No results found.". Is there a solution to handle this case as well? Thanks a lot.

[rjthibod]

Yes, you can use append to include a dummy value that gets filtered out if there is real data.

| tstats count where index="abc" by _time span=1h 
| addtotals fieldname=Total
| append [|gentimes start=-1 | addinfo | table info_min_time | bin info_min_time as _time span=1h | eval count = 0 | eval Total = 0 | table _time count Total]
| where (Total = 0 OR count > 0)
| timechart span=1h max(count) as count
| fillnull value=0

[jsh315]

That worked. Thanks a lot!