How to typecast timestamp_ntz to CST format using Spark/PySpark while writing to Snowflake - apache-spark

Question

As we know we have 3 timestamps in Snowflake,
TIMESTAMP_NTZ
TIMESTAMP_LTZ
TIMESTAMP_TZ
So while writing timestamp to snowflake table, it bydefault takes as TIMESTAMP_NTZ.
How snowflake can take the timestamp in CST timezone while writing to snowflake table?

Answer 1

First it's important to know what timezone has been set as the default for your account/session:
SHOW PARAMETERS LIKE 'TIMEZONE';
Then change the default for your session to CST
ALTER SESSION SET TIMEZONE = 'America/Chicago';
And thereafter any selects of current_timestamp will be providing the data in the right timezone
SELECT CURRENT_TIMESTAMP;
This is a great article for reference:
Snowflake timestamp datatype ref
Assuming you have control over the precise column type in your table, I found that TIMESTAMP_TZ is how you want to define your table. Here's working example of everything I did:
alter session set timezone = 'America/Los_Angeles';
create or replace table ts_test(rn number, ts timestamp_tz);
insert into ts_test values(1, current_timestamp());
insert into ts_test values(2, '2019-12-10 07:50:00 -06:00');
insert into ts_test values(3, CONVERT_TIMEZONE('America/Chicago', CURRENT_TIMESTAMP()));
select * from ts_test;
if the timestamp is being generated in code, then make sure you include the UTC offset when inserting (rn 2). If you're using the current_timestamp() which is LA, then make sure you convert to CST (rn 3).
if the table is being generated and you don't control the timezone default, then issue this first:
alter session set timestamp_type_mapping = timestamp_tz;