The DSN-less connection string below FINALLY worked for me, in windows 10. I created a file DSN, then copy/pasted the string into the python code, as a template.   Three lessons that I learned from this struggle:   1) kerberos is CASE SENSITIVE. Your kerberos realm in the string MUST be uppercase. 2) The Cloudera driver doesn't like spaces in between the semicolons in the string. Avoid them. 3) If you don't need connection pooling, turn it off with a pyodbc.pooling = False statement. import pyodbc strFileDSNAsAstring = "DRIVER=Cloudera ODBC Driver for Apache Hive;USEUNICODESQLCHARACTERTYPES=1; \ SSL=0;SERVICEPRINCIPALCANONICALIZATION=0;SERVICEDISCOVERYMODE=0;SCHEMA=database;PORT=port; \ KRBSERVICENAME=hive;KRBREALM=uppercaserealm;KRBHOSTFQDN=hostfqdndomain;INVALIDSESSIONAUTORECOVER=1; \ HOST=host;HIVESERVERTYPE=2;GETTABLESWITHQUERY=0;ENABLETEMPTABLE=0;DESCRIPTION=Hive; \ DELEGATEKRBCREDS=0;AUTHMECH=1;ASYNCEXECPOLLINTERVAL=100;APPLYSSPWITHQUERIES=1;CAIssuedCertNamesMismatch=1;" try: pyodbc.pooling = False conn = pyodbc.connect(strFileDSNAsAstring, autocommit=True) except: print("failure.") else: conn.close() print("success.") ... View more