SAS might crash when the following are true:

• You use SAS/ACCESS^® Interface to Spark with PLATFORM=DATABRICKS.
• You use BULKUNLOAD=YES to read data from Databricks.
• Databricks contains a column that is 32,767 or more characters in length.

If all of the above are true, SAS might crash. If you capture the SAS log, errors similar to the following occur:

ERROR: The size of the data exceeded the maximum allocated buffer size for BULKUNLOAD.
ERROR: read_and_convert_data_file(1): get_column_data failed, rc = 0xfffffc1b

Workaround

Method 1: Use BULKUNLOAD=NO

Unless data volumes are small, this method might significantly affect performance.

If you use a client such as SAS^® Enterprise Guide^®, and you attempt to open a problem table, you must use this method.

Method 2: Use either the DROP= or KEEP= data set options to exclude the problem column