Fail to publish GP as a service

Question

I have no idea what the issue is here, I created this scripting tool which works fine when I run it through ArcGIS desktop. However, for some reason, I am unable to publish the tool as a service.The following is the error (00178) that I keep receiving:

"Data: \, used by script cannot be copied to the server". The script is very simple as it is just doing a simple calculation derived from a feature class which is currently stored on the SDE geodatabase. I've already looked through the documentation and none of the solution seems to resolve my issue.

    import arcpy, csv, datetime

    output = arcpy.GetParameterAsText(0)
    DATA = r'C:\!PROD_Connections_SDE\PROD@ABC.sde\ABC.PROD.S2_testbuild\ABC.PROD.ABC_testline'

    #table = arcpy.MakeFeatureLayer_management(DATA,"data")

    # Map of UTM Zone WKIDs, modify as needed
    utmZones = {10: 32610,
        11: 32611,
        12: 32612,
        13: 32613}

        pLength = 0.0
        hLength = 0.0

       def getUTMZone(geometry):
       #http://gis.stackexchange.com/questions/13291/computing-utm-zone-from-lat-long-point
           from math import floor
           geo = geometry.projectAs(arcpy.SpatialReference(4326))
           lon = geo.trueCentroid.X
           return int(floor((lon + 180)/6) + 1)

       with arcpy.da.SearchCursor(DATA,["SHAPE@","Activity"]) as cur:
           for row in cur:
               utmZone = getUTMZone(row[0])
               sr = arcpy.SpatialReference(utmZones[utmZone])

               # Project line to correct UTM zone and add its length to the correct sum
               geom = row[0].projectAs(sr)
               if row[1] == "HDD":
                   hLength += geom.length
               else:
                   pLength += geom.length

       # Write data to CSV
       with open(output,'wb') as outFile:
           date = datetime.datetime.today()
           headers = ("Date","P Total Length (m)","H Total Length (m)")
           record = (str(date),str(plength),str(hlength))
           writer = csv.writer(outFile)
           writer.writerow(headers)
           writer.writerow(record)

Answer 1

The analyser likes to pick out random items and claim the script is using them, for example this post on Geonet.

I got round the problem by putting the SDE connection file on a network share that the server could also see and referencing it with a full UNC path:

INWorkspace = r"\\store\UTILITIES\SDE Connections\xxx_geodb.sde"