Possible athenad memory leak

[sshane]

It was using 15% of memory, causing low memory events. I had not rebooted the device in a few days iirc and had uploaded many rlogs.

I tried to attach with memray, but instead the process just crashed so I didn't get any information about this. @deanlee could this be similar to the leak you found in athena previously?

[deanlee]

It's hard to identify the exact issue, as the code is unclear and difficult to understand. However, multiple factors could be causing the resource leak. One possibility is that the athenad's implementation might cause garbage collector to accumulate objects over time, leading to memory issues during long runs.
additionally, thread race conditions related to queue and other global variable (like cur_upload_items, cancelled_uploads,...) usage, as well as bad practices like not calling task_done() after queue.get(), could also contribute to issues.