Hello all,
I followed the steps in sample.sh and successfully built my first cube.
Now I try to understand the procedure behind it. But there are some
files which were not created within the sample.sh, e.g. metadata.
My aim is to create a cube programmatically without the Web UI. This is
the stage of affairs with some questions:
The first thing sample.sh does is loading tables in hive. I have some
.csv files and created an executable hiveql-file for creating the
schema. But is there a class in kylin (or hive) which can produce a
hiveql-file from .csv automatically?
In step two sample.sh uploads metadata in hbase. But how to create these
metadata - json files? I know there were plenty of *Desc-classes like
CubeDesc, DataModelDesc, DimensionDesc and so on. And with
JsonUtil.writeAsIndentString() I can convert my java object in json
format. But some json-files refer to each other. Is there a class which
creates all the metadata in one place?
Do I need the json-files or is it enough to create a CubeDesc and then
calling cubeMgr.createCube("myCube", "testProject", cubeDesc, null)?
Thanks in advance,
Marcel Jacob.
