CrushFTP v11 now has a separate engine for running jobs. You can control its memory for this JVM via the prefs.XML and changing "jobs_memory_param" to an integer number representing MB to use. 2048 or 4096 for example.
The main CrushFTP server engine automatically launches this JVM at startup and kills it at restart or shutdown. Its handled for you.

If you want to manually manage this JVM, it's possible to run the jobs engine on its own.

java -Dcrushftp.jobs_location=./ --add-opens=jdk.management/com.sun.management.internal=ALL-UNNAMED -jar plugins/lib/CrushFTPJarProxy.jar -JOB_BROKER 2500

See the attached script file for Linux: attachment Installing via the script:

./crushjobs_init.sh install

The prefs.XML must have the parameter configured for the port for the jobs engine:

jobs_host_param (default is 127.0.0.1:0...so for this, set it to 127.0.0.1:2500)
job_broker_key (default is RANDOM and should stay this way) 

"job_broker_key" is a unique key so CrushFTP and the jobs engine know they are talking to the right one. It should stay set as RANDOM unless you plan to run the jobs engine on a separate machine and do not want it restarted if the main CrushFTP server is restarted.

---

Its also possible to run a pool of job engines. RemoteJobsEngine