feat(ingest/mssql): improve docs on using odbc (datahub-project#10370)

Co-authored-by: Jeff Lewis <[email protected]>
Co-authored-by: Harshal Sheth <[email protected]>

Author: 3 people

metadata-ingestion/setup.py

@@ -233,6 +233,11 @@
     *pydantic_no_v2,
 }
 
+mssql_common = {
+    "sqlalchemy-pytds>=0.3",
+    "pyOpenSSL",
+}
+
 postgres_common = {
     "psycopg2-binary",
     "GeoAlchemy2",
@@ -371,12 +376,8 @@
     },
     "mode": {"requests", "tenacity>=8.0.1"} | sqllineage_lib | sqlglot_lib,
     "mongodb": {"pymongo[srv]>=3.11", "packaging"},
-    "mssql": sql_common
-    | {
-        "sqlalchemy-pytds>=0.3",
-        "pyOpenSSL",
-    },
-    "mssql-odbc": sql_common | {"pyodbc"},
+    "mssql": sql_common | mssql_common,
+    "mssql-odbc": sql_common | mssql_common | {"pyodbc"},
     "mysql": mysql,
     # mariadb should have same dependency as mysql
     "mariadb": sql_common | {"pymysql>=1.0.2"},

metadata-ingestion/src/datahub/ingestion/source/sql/mssql/source.py

@@ -151,6 +151,7 @@ class SQLServerSource(SQLAlchemySource):
     - Column types associated with each table/view
     - Table, row, and column statistics via optional SQL profiling
     We have two options for the underlying library used to connect to SQL Server: (1) [python-tds](https://github.com/denisenkom/pytds) and (2) [pyodbc](https://github.com/mkleehammer/pyodbc). The TDS library is pure Python and hence easier to install.
+    If you do use pyodbc, make sure to change the source type from `mssql` to `mssql-odbc` so that we pull in the right set of dependencies. This will be needed in most cases where encryption is required, such as managed SQL Server services in Azure.
     """
 
     def __init__(self, config: SQLServerConfig, ctx: PipelineContext):