Deepstack Logs? OR timeout / error 100 solution?

I’m trying to troubleshoot some DeepStack / Blue Iris problems. I’m not sure where the issue is right now, and it would help if I had more insight into what DeepStack is doing. I searched the documentation and the forums and the closest thing I could find is this thread from 2019 saying that logs were a priority.

Does DeepStack write a log file anywhere, or can it be made to (specifically when used with Blue Iris)?

The problem I’m having is that a lot of my alerts recently are timing out or receiving an “error 100” (which I believe is the same thing). Some of these seem legitimate, e.g. BI is set to timeout after 15 seconds, most of them are right at around 15,000ms. However, a large number of them are shorter periods of time (100’s of ms). Here’s a screenshot from the BI logs.

Things seem fine for a half hour or so after a reboot. Analysis times seem to be 100ms or so, objects are being recognized, etc. However, performance quickly degrades.

It almost seems like a memory leak or something, but memory/cpu/gpu utilization seem fine, and I"m not even sure that’s a thing with DeepStack. If the Blue Iris logs are right though, DeepStack is being restarted, but that isn’t helping, so something else is going on. Again, it might help if I could see into what DeepStack is going, preferably with a flat log file I can ingest into Splunk. I might be able to find a trend (performance drops off a cliff after X number of API calls?).

This is on a physical host, running DeepStack GPU on an Quadro P400, core i7-10700, 32 GB of ram.