Hi, According to this link <http://kylin.apache.org/docs/howto/howto_optimize_build.html>, “Too high cardinality is not suitable for dictionary”. Have you tried other encoding methods such as “fixed_length” for the UHC column? I am not sure maybe this is not the best solution but it works for me. You can also check "Deal with Ultra-High-Cardinality Columns" section in this page <http://kylin.apache.org/docs/install/configuration.html>.
On Wed, May 6, 2020 at 4:15 PM 我是天才 <[email protected]> wrote: > Hello: > I have a table and it has three columns,it has 5736604 rows,the three > columns distinct value counts is 1843696 5736485 1742356.I build cube and > got a error on 3rd step,it said:"Error: Java heap space",the step name is > " Extract Fact Table Distinct Columns". > Can you help me? thanks! >
