Hi Everyone.
Currently, the process of loading CSV data into Vertica is as follows
# csv_converter: Module that performs format conversion of CSV files (delimiters, end-of-line characters)
# load_csv.py: Load data using COPY command (python)
cat ${INPUT_CSV_FILE} | csv_converter | load_csv.py
At this time, if an error occurs in csv_converter, load with the COPY command.
I want to force termination and rollback.
With the current method, if an invalid record is found in csv_converter
The module will terminate, but normal records prior to that will be loaded into Vertica.
Since we want to have all records registered or not registered in each CSV,
we need to make sure that all the records are loaded into Vertica.
The above behavior is undesirable.
If anyone knows of a better way to do this, we would be very interested in hearing about it.
That's all, thank you for reading.