Separate Template for the Two component MR Jobs of Batch Ingestion


I am using Hadoop-based Batch Ingestion for ingesting data for my use case. This job has 2 component MR jobs - one which determines number of partitions and the second one which generates indexes.

Now, both of them use a common config template. But for my use case, I have observed that the determine partitions job uses about 400-500 MB of reducer memory whereas the second job uses about 3.5 GB of reducer Memory.

Since there is a common config, a lot of reducer memory is wasted since is is specified to be 4 GB in the common config.

Is there a way to specify two different configs or at least have this property specified differently for each of the jobs?