What you could do to possibly find the leak is remove part of your network while testing: maybe try with the cacheTOP first (what’s the current cache size ?), how is the live stream received ? Video Stream In ? What parameters?
If you look at the GPU memory used by the process (details view of Task Manager, add the GPU memory column), do you see it continuously increasing over time ? After 10 minutes… 1 hour… 4 hours ?
Are there any recommendations/do’s&don’ts on running a touchdesigner project for long periods? (as the plan is to let this project run for 24/7…)
Usually on production projects running 24/7, some developers would schedule a reboot of the machine at some point, or restart of the TouchDesigner process at a time X, using various process managers, such as NodeJS’ PM2.
That being said, it might “hide the leak” in a way where the leak would still be happening but not fill all the RAM or GPU Memory before reboot / restart of process… It doesn’t mean that the leak is not here and that it shouldn’t be fixed.