I have a Desktop Script tool that works fine when run on desktop, but does not work when the results are turned into a geoprocessing serice. Running ArcGIS 10.2 on Desktop and 10.2 on server.
I have found that the geoprocessing service fails at the line of code where it tries to to load a data file I have saved to file. I have tried saving the data files using pickle, json, and also array files as .npy and .npz. It doesn't seem to make any difference. The geoprocessing service just will not load any array type file I provide. I have tried letting the geoprocessing packager save the file to the server, I have saved the files to a registered folder, and it doesn't seem to matter. Even text files will not load.
Has anyone been successful at loading a numpy array, dictionary, or similar type file into a geoprocessing service on arcgis server? If so, how did you do it?
I added error handling to my script, and here are the messages when I run the Geoprocessing Service:
Traceback Info: File "E:\ARCGIS~2\DIRECT~1\ARCGIS~4\ARCGIS~2\Grace\HT5C33~1.GPS\EXTRAC~1\v101\HTAT_S~1\140307_HTAT_debug.py, line 74, in [module] npzfile = numpy.load(array_package) Error Info: [Erno 2] No such file or directory: u'C:\\Users\\arcgis\\AppData\\Local\\Temp\\..\\cd\\tooldata\\package131123.npz
The E: drive is on the gis server machine and the .npz file is my data file I am trying to load. 140307_HTAT_debug.py is my source script. Apparently the npz file was copied over to the server, but not sure what that path means with the double dots, and why the server can't read it.