Expose usedPreaggregation through REST load API in prod mode

[flochaz]

Is your feature request related to a problem? Please describe.
We are heavily relying on pre-aggregation and relying on job API to force rebuilt of pre aggregation.
The API support cubes or preAggregation parameter but there is no way to know what cube or preAggregation is used by a given query.

Describe the solution you'd like
get usedPreaggregation detailed in prod mode

Describe alternatives you've considered
The naive alternative was to try to deduct the cube targeted using the dimension of the query but it is to reliable and force us to refresh the entire set of pre aggregation with in a cube.

[igorlukanin]

@flochaz Thanks for filing this! Just one more time, could you please explain, in the simplest terms possible, what are you trying to achieve here? Run a query, understand if it targeted a pre-aggregation, then refresh that pre-aggregation via the Orchestration API? Is this a correct guess? If so, why would you like to do it this way?

[flochaz]

Queries to create the pre aggregations are quite complex in our case and take db loads quite heavily so we limit the auto refresh to every few hours but sometime client want the latest data so we offer as an extra option to force the refresh for a specific graph/query. I hope it clarify things. But maybe there is other way to achieve this though

[igorlukanin]

@flochaz Please take a look at this comment: #9584 (comment)

Would it be of help to you?

[flochaz]

@igorlukanin, kind of but I still find it weird to have to guess the preaggregation used (and its name) by a query to be able to use fully the Job API (using preAggregations parameter) ...
I haven't tried yet the renew option but I don't really need to be in foreground since my preAggregation can take hours ...

[igorlukanin]

but sometime client want the latest data so we offer as an extra option to force the refresh for a specific graph/query

but I don't really need to be in foreground since my preAggregation can take hours...

I'm sorry, just trying to understand how this is supposed to work together. So, you'd like to provide the user with an option to refresh a specific graph/query but not in the foreground—so how are they supposed to know if the data is finally refreshed?

Really trying to nail a solution that works for this use case.