I've been able to successfully query and plot documents with geo_points in maps in Kibana alongside time series plots of other data in standard fashion. When I select the time range in a time series plot, it correctly narrows the geo_points displayed in the map. However, I would like to be able to do a shape query to select the time range displayed as well. This does not happen naturally.
In fact all of my time series data becomes de-selected when selecting any part of the map data which matches in time. Do I need to have geo_point data fields named the same in my other documents? If not is there an aggregation that I could do to accomplish this?
I may simply be using data from different indexes which don't link. I just don't understand why geo-query select a set of geographical points (which have timestamps associated with them) and it doesn't narrow the time range and in fact de-selects everything when I do the geographical query.
So when I select a shape on a map, it will filter the view to any document with the same field geo_point which lies within the bounds? If so that makes sense.
There's a broader theme I'm looking to figure out: How can I automatically adjust the time range according to a KQL or Lucene query? Is there a setting that might cause Kibana to automatically adjust the view time range resulting from a selection query like survey_name: "my_specific_survey"?
I realize this -- in general -- isn't practical. The range could be massive. But what if our selections are known to be reasonably small and we would like to automatically adjust the time range resulting from a query. Is that possible?
That's exactly the defition of what the shape filter does. One thing that helps a lot to understand how things work is to check the Inspector tool that displays the exact queries that are being passed to Elasticsearch.
So for example using the Kibana Flights sample dataset added as a Document layer, and with the DestLocation as the geometry rendered (remember an index can have multiple geometry fields!) the bounds filter generates a query like this:
Forgot to mention that every layer can ignore the global search bar and date picker individually, which can be convenient in certain cases where we want our data to render independently (or in conjunction with other regular datasets). Layers can also have their own filter as well.