Memory leak while using in_memory ArcGIS 10.1

Question

I am running a Python script to make point feature layer from a series of netCDF files and then perform a spatial join with a number of polygons and save the results as a table. I am using arcpy.MakeNetCDFFeatureLayer_md and address its output to a variable in 'in_memory' workspace. My problem is that even though I am deleting the 'in_memory' after each loop using arcpy.Delete_management('in_memory'), it does not free up my RAM and after a couple of iterations I run into Memory Error. The function that is being called for each netCDF file looks like this:

def spatialJoin(nc):
 arcpy.env.overwriteOutput = True
 arcpy.env.workspace = 'in_memory'
 arcpy.MakeNetCDFFeatureLayer_md(nc, "SM", "lon", "lat", "SMPoints", "time;lat;lon", "", "", "", "BY_VALUE")
 <does the spatial join>
 print arcpy.Exists('SMPoints')
 arcpy.Delete_management('SMPoints')
 print arcpy.Exists('SMPoints')

And the console prints:

True
False 

for each iteration, which means Delete_management is working. But checking the python process in windows task manager shows that the data is piling up in RAM! I have tried deleting individual features inside 'in_memory' as well as deleting the whole 'in_memory' but nothing seems to work. Am I missing something here?

P.S. Spatial join does not affect the problem here, problem exists even when I completely comment out the spatial join section.

• JamesLeversha: It's nasty, but you could try and delete the arcpy module import after each iteration then re import it.
• fatih_dur: Did you see gis.stackexchange.com/questions/19684/…? Also it's worthwhile to have a look at geeohspatial.blogspot.com.au/2013/12/…
• Monobakht: Thanks @fatih_dur, the second link was useful. I managed to free up the memory by using multiprocessing module. It slowed down the process a bit, but the memory leak has stopped ;)

Answer 1

You shouldn't need to set your environment workspace to in_memory as you are doing in your script, so you can drop this line:

arcpy.env.workspace = 'in_memory'

Also, ensure that you are using double-quotes when referencing your in_memory workspace inside of the geoprocessing tool.

arcpy.Delete_management("in_memory")

• Midavalo: "ensure that you are using double-quotes when referencing your in_memory workspace" - that's a new one for me, can you explain/reference the double-quotes requirement over single-quotes?
• Monobakht: Thanks @Radar, but changing to double-quote didn't help!