Elasticsearch: In Explore we add unnecessary `date_histogram` aggregation when requesting logs

[fmassot]

What went wrong?

What happened:

On the explore view, when I sent a request of query type "logs", the backend builds a query with a date_histogram aggregation. This aggregation is useless because the frontend is also firing a logs volume request.

What did you expect to happen:

The backend does not need to build a logs query with a date_histogram aggregation. This means that we can remove all those lines

How do we reproduce it?

• Open Grafana and go to the explore view
• Select an Elasticsearch datasource
• Select the "logs" query type
• Launch a query

What Grafana version are you using?

10.1.0-116348pre-ubuntu

Optional Questions:

Is the bug inside a Dashboard Panel?

No response

Grafana Platform?

Docker

User's OS?

macOS Monterey

User's Browser?

Google Chrome Version 112.0.5615.137 (Official Build) (arm64)

Is this a Regression?

None

Are Datasources involved?

Elasticsearch

Anything else to add?

No response

[ivanahuckova]

Thank you for creating this. You are right, but the possible issue in removing this is in dashboard queries - because there you can use time series panel to display this data. So this might be breaking change. We need to think this trough because there are couple of different options on how to solve this:

• add a param and not add aggregation in Explore
• deprecate first and then add a breaking change

So at this point, we won't accept PRs tackling this as we need to reach consensus within the squad on what we want to do.