ArcGIS Pro 3.0.2
Raster functions are great. Instead of taking up to 36 hours to calculate NDVI, it takes but a few seconds. However, the down side is that further analysis on the raster function result is completely spurious. Sometimes it works, sometimes not. And there is no rhyme or reason why.
I'm using AGP in a virtual desktop with 32 GB ram, terabytes of disk space and a cpu somewhere between 3 and 4 GHz.
The result of the raster function (in this case) is a raster with about 300,000 rows and columns and a resolution of 16 cm. It has no pyramid layers but everything looks good when I zoom up to a zone. I'm 100% sure that all of the zones overlap the raster function raster (I designed it this way). I'm using six different zonal regions (converted to 2m rasters).
Running Zonal Statistics as Table results (sometimes) in a table with Mean values for the raster. Running the same exact process again - for example on a new instance of ArcGIS Pro with the same inputs (checked to make sure they exist and overlap) - results in empty tables.
This happens with the batch function. It happens using the normal tool (no batch). It happens unpredictably.
Sometimes I get results. Sometimes not. Same dataset. Same virtual desktop. No other things running in the background.
Is this a known bug?
Tech Support is your best option especially if you want to check against published and unpublished known bugs
Normally, I would agree with you but ESRI Tech Support is sort of a joke where I live. First you send in the question, then it's translated to English by the local ESRI people. Then the question is sent to Redlands (nine hour time difference). Then a day or two goes by. Then a response is translated from English to the local language. In any case, the response from Redlands is typically: can you send us the data (which is almost always impossible) .
In fact, the local ESRI people often come to my company for solutions. However, I find that the various forums offer the best help and that ESRI Tech Support leaves a lot to be desired.