Layer probe, random requests or same tile

[jorgejesus]

Hi everyone,

I have a question/discussion concerning GeoHealthCheck implementation.

When we have a WMS service and use a layer probe (GetMap), will it always use a specific bbox/tile or does it do random requests? The "problem" is that if we have constant requests to a specific tile, we will have lower response times as the tile will likely be cached. More realistic monitoring would use random requests each time and then maybe aggregate statistics (average, standard deviation).

Is this random request approach implemented in GHC?

Thanks.

[justb4]

Good question! Answer is: no (not yet).

The WMS probe is here: https://github.com/geopython/GeoHealthCheck/blob/master/GeoHealthCheck/plugins/probe/wms.py
There is also a WMS Drilldown Probe that traverses a WMS from its Metadata, starting at GetCapabilities.
The WMTS Probe has some randomness built-in: https://github.com/geopython/GeoHealthCheck/blob/master/GeoHealthCheck/plugins/probe/wmts.py

But a good idea, though with some possible focus points:

• a random bounding box may "overask" the WMS and underlying DB, become a sort of DoS. But then any robust WMS should be resistant.
• a random width/height may be outside configured WMS limits (viz. MapServer).

But randomness within boundaries, and possibly if the WMS is based on a tileserver like MapProxy: very doable. PR welcome!