Justin Su

I'd instead add a note after line 60. Something like: > You might need to do this for standard library modules, as the standalone binaries only include a minimal Python...

Upon further thought, this probably isn't a regression. I hadn't noticed any SIGKILLs before, but I think I'm just consistently getting retryable errors in certain regions. EDIT: After adding more...

I've already provided a description of the issue and a reproduction. What are you looking for, specifically? You can build the script and run it yourself to see. Use a...

I used btop. But you can also run it until it gets killed by OOM.

`ps aux` gives me these results. Old process: ``` USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND ec2-user 365383 0.6 41.9 2340868 791440 pts/4 Sl+ 14:05 2:31...

In the `ps aux` snippets I added earlier, both %MEM and RSS have gone up significantly when comparing the old process (running for 2.5hr) vs the new one.

Okay, let me try to narrow down some variables. One thing I can think of is that I'm building on a c8g.large and running on a t4g.small. @wty-Bryant Can you...

I might not be able to set up pprof, since it seems a bit more involved than I expected after a quick look. But I'll try to get a simpler...

Thanks! I'll find some time later to try pprof.

2 profiles captured ~8 minutes apart (requests were made with 0.5s sleep interval): [pprof.ec2-run-instances.alloc_objects.alloc_space.inuse_objects.inuse_space.001.pb.gz](https://github.com/user-attachments/files/24155653/pprof.ec2-run-instances.alloc_objects.alloc_space.inuse_objects.inuse_space.001.pb.gz) [pprof.ec2-run-instances.alloc_objects.alloc_space.inuse_objects.inuse_space.003.pb.gz](https://github.com/user-attachments/files/24155866/pprof.ec2-run-instances.alloc_objects.alloc_space.inuse_objects.inuse_space.003.pb.gz) And one more profile captured ~30 minutes later. Looks like memory usage went _down_ because...