Netezza - nzload data with default values - sql

Question

I am trying to load int data into a table on a Netezza server. However, I do not know how to load in data with a default value in the case that it does not exist or is null.
Right now, the table consists of two columns with both having its own default value.
Attribute | Type | Modifier | Default Value
----------+---------+----------+----------------
number1 | integer | | 0
number2 | integer | | 100
I am currently running the following nzload command: nzload -cf test.log
The test.log file looks like this:
DATAFILE /usr/me/test.dat
{
Database test
TableName numberTest
Delimiter '|'
}
The test.dat file looks like this:
1|2
3|4
5|6
7|
|8
The issue I am faced with is that while the command runs fine, the integer values default to Null as opposed to what the default value was set to. I have tried using insert within the nzsql, and that creates the correct default values, but I was wondering if there is a way to do this with nzload.
Any help would be much appreciated.

Answer 1

The default value constraint will be enforced when performing inserts where the column with the default value is not referenced in the column list for the insert.
For example:
TESTDB.ADMIN(ADMIN)=> create table default_test (col1 varchar(10),
TESTDB.ADMIN(ADMIN)(> col2 varchar(10) default 'myDefault', col3 varchar(10));
CREATE TABLE
TESTDB.ADMIN(ADMIN)=> insert into default_test (col1, col3) values ('A','C');
INSERT 0 1
TESTDB.ADMIN(ADMIN)=> select * from default_test;
COL1 | COL2 | COL3
------+-----------+------
A | myDefault | C
(1 row)
However, when you are performing an nzload, Netezza is actually performing an insert into the target table with a select from an external table defined on your load datafile. In doing so it is including each column in the column list, and therefore the default value will not be triggered, even if the value in the external table's data file is NULL or an empty string.
[nz#netezza test]$ cat test.txt
A,B,C
D,,F
G,NULL,I
TESTDB.ADMIN(ADMIN)=> create external table default_test_ext
TESTDB.ADMIN(ADMIN)-> sameas default_test using (
TESTDB.ADMIN(ADMIN)(> dataobject '/export/home/nz/test/test.txt' delimiter ','
TESTDB.ADMIN(ADMIN)(> );
CREATE EXTERNAL TABLE
TESTDB.ADMIN(ADMIN)=> select * from default_test_ext;
COL1 | COL2 | COL3
------+------+------
A | B | C
D | | F
G | | I
(3 rows)
TESTDB.ADMIN(ADMIN)=> select * from default_test_ext where
TESTDB.ADMIN(ADMIN)-> (col2 is null or col2 = '');
COL1 | COL2 | COL3
------+------+------
D | | F
G | | I
(2 rows)
Since NULL and empty strings are valid values, and nzload is referencing that column in its insert the default value cannot/should not be used. It's working as I would expect it to, however it would definitely be useful if you could tell nzload to transform NULLs or empty strings to the default value for a column. Unfortunately that functionality doesn't currently exist (at least not to my knowledge).
While this is hyper-kludgey, I have gotten around this for data loads by doing the external table manually, and loading in two steps.
TESTDB.ADMIN(ADMIN)=> truncate table default_test;
TRUNCATE TABLE
TESTDB.ADMIN(ADMIN)=> insert into default_test (col1, col3)
TESTDB.ADMIN(ADMIN)-> select col1, col3 from default_test_ext
TESTDB.ADMIN(ADMIN)-> where (col2 is null or col2 = '');
INSERT 0 2
TESTDB.ADMIN(ADMIN)=> select * from default_test;
COL1 | COL2 | COL3
------+-----------+------
D | myDefault | F
G | myDefault | I
(2 rows)
TESTDB.ADMIN(ADMIN)=> insert into default_test
TESTDB.ADMIN(ADMIN)-> select * from default_test_ext
TESTDB.ADMIN(ADMIN)-> where (col2 is not null and col2 <> '');
INSERT 0 1
TESTDB.ADMIN(ADMIN)=> select * from default_test;
COL1 | COL2 | COL3
------+-----------+------
A | B | C
D | myDefault | F
G | myDefault | I
(3 rows)

Answer 2

Netezza does not enforce the default value constraint. It merely exists for notation. IBM Documentation. In order to fix your table you must run update statements.