In our project, we are running many kettle jobs and transformations in parallel.
We got a OOM error during some huge data load for some specific job alone.

we thought of keeping multiple kettle.sh with different Xmx configuration in such a way that the job taking huge load will be triggered using kettle.sh with max Xmx.
we usually use pan.sh to execute any single transformations as you may aware...

As there are lot of transformations running inside the job, do they take pan.sh available in the DI to execute them or will it be executed by the kettle.sh itself ?
Do we need to maintain separate pan.sh as well similar to kettle.sh ? if so how can we do it ?

kindly help