Partition in Join stage

[G SHIVARANJANI]

Hi,


Could any one explain what does partition do in join stage,

how can i disable the auto match in mappings , so that i can send :

say col 1:country into col b:surname .

[ray.wurlod]

Proper key-based partitioning makes sure that, when a key value comes in from the left input, it can be sure that all occurrences of that key from the right input are on the same node as the value on the left input.

There is no auto-match in mappings in the Join stage. It just appears that way. You can override any mapping in the GUI on the Mapping tab for the Output link.

[G SHIVARANJANI]

Hi ,

if the mapping is this way


EXP------col_nm---- derivation---- colmnname

inp.col1--col1______ inp.col1------ col1
inp.col2--col2______ inp.col3------ col2
inp.col3--col3______ inp.col3------ col3


why am i getting inp.col2 in col2 of output when i traced out the result....

instead of getting inp.col3 in col2 of output.


its automatically putting the data as per the name of columns..
[/img][/url]

[ray.wurlod]

What it puts is a default. Change it if you want it different.