I recently wrote a python script to take CAMA data from an Oracle database in the Assessors' office and update a table on our enterprise geodatabase; basically just taking property ownership information from one table and updating it daily in another so we can consume it in our GIS system. At first, I tested on my desktop machine where it ran just fine. However, when I went to test it on our GIS server, that's where the trouble began. It seems to not be able to connect to the Oracle database via the script, as I presented with the 'Error: 000732: < dataset name > does not exist or is not supported' error when running any tool that requires a connection as an input (e.g MakeTableView, CopyRows, TableToGeodatabase, etc). Definitely has me stumped because I am able to connect to the database manually via an OLE DB connection in both ArcCatalog and in ArcMap; however, when run in IDLE it fails.
At first I though it might be a 32-bit vs 64-bit issue. The desktop machine uses the 32-bit flavor of python in IDLE. Thought perhaps the server was forcing the script to run using the 64-bit, but after running this helpful little bit of code:
it appears as though the server is running 32-bit as well.
As far as I know the Oracle database we are connecting to is 11g. However, both the server and desktop machines are using the 12c Release 2 (188.8.131.52.0) 32-bit flavor of the Oracle client because the 11g client does not does support Windows 10. Currently, my desktop machine is running 10.3.1 and the server machine is running 10.4.1, w/ python versions 2.7.8 and 2.7.10, respectively.
The only thing I haven't tested is installing the 64-bit version of the client software on the server side. Even though IDLE tells me it's running on the 32-bit side of things, I have a suspicion that somehow the connection to the database needs a 64-bit component ...