Geomap panel truncating lat/lon data elasticsearch


Grafana v8.4.5



Trying to plot IoT GPS data into geojson layer (indoors, so need more accuracy)

The data is written to an index in elasticsearch. Data type for lat/lon fields is float.

i.e. Querying ES CLI, data looks like this.

“nodes_longitude” : 9.3727508,
“nodes_latitude” : 49.227037,

Geomap Configuration

The query:
Metric: Logs (Other metrics would not return the individual fields for lat and lon)
Transformation: Filter by name
Standard Options: Unit: Short, Decimals: 7
Overrides for both fields: Unit: Short, Decimals: 7
The Marker layer is configured as Coords using the above two fields.

Accuracy is incorrect, because Grafana seems to be truncating the decimal values. See Above Table view of the geomap panel.

Any help with this is highly appreciated. Only found two similar topics, but they do not seem to have an answer, and I tried the suggestions within the topic without luck.

Use query inspector to see what is received from ES actually. Logs for Geomap panel doesn’t look like a good metric. I would use Raw Data + Misc/none for unit options.

1 Like

The query inspector shows the data correctly with 7 decimal digits. I tried Raw data and None Unit. But it still looks the same.
The only way I can see the number properly (or close to the real value), is if I do a “Configure Query Result” transformation, and set those two fields to decimals. But that adds a longer decimal value to the field and every other numerical field, even if it is just an integer (which is incorrect)

In kibana what is the data type of those fields?

They are float type, but I think I found the issue. I seem to need a zoom level of about 26 to see the points properly in my geojson layer, and the plug-in seems to limit the render to about zoom 18. The table view is just rounding the values, but if I zoom in all the way, I can see they are distributed. Just not rendered at the right coordinates.

Is just strange that the geojson layer renders correctly, but not the heatmap/marker points.

1 Like