SL memory utilization

I have created a pipeline which deals 2 GB data. When I manually terminate the pipeline or if its fails, pipeline stops. But SL jvm is still holding on a huge chuck of memory causing snaplex crashes if I run the pipeline again.

My current work around is restarting the entire snaplex which is fine in development but horrible solution for production. Has any one encountered this issue or suggest me a better solution?

Thank you in advance,

I would recommend checking with Support.

Scared of doing that but I will try