Hello,

I am having a real pain reducing as much memory as I can in my PDG network. I'm trying to file cache each asset at the start as the rest of the process uses barely any memory afterwards. I've even gone as far as to create a service specifically to force only one work item be ran for the ROP Fetch to avoid multiple work items running in parreler at that specific stage. However as it doesn't restart hython after each work item so the memory accumulates and eventually runs out. Is there anyway to force hython to restart and intern dump the memory? I know it will be a bit slower needing to restart the scene file every time but it's a compromise that I'm fine with.

Services record their memory usage each time a job finishes as part of the response sent back to Houdini. We don't use a finer granularity than that as it would require more active polling of the process while a job is in flight.

It's possible to set up automatic resetting or restarting of clients when that measured memory usage grows too large after a job finishes. "Reset" in means a special message is sent to the worker asking it to unload assets/its scene, and "Restart" is a full tear-down/restart of the worker process.

The settings for that are available in the service panel or on the Service Create node, if you're using the node-based workflow for managing your services. I think that'll give you what you need in this case.