Disk space exhaustion when leaving Codex open

I’ve been a Codex user since the early days of it coming out, but I must say, ever since a few versions ago, I have been dealing with a nasty bug where the SQLite logging module inside Codex emits so many trace events that disk space gets filled constantly.

I am talking about this: github (dot) com (slash) openai/codex/issues/22444

Here is a more recent Github issue by another person: github (dot) com (slash) openai/codex/issues/28224

Honestly, I can reproduce this bug so many times that I even made a Bash script that cleans up the SQLite WAL file and automatically terminates the Codex processes to release the file. Why is this bug being ignored for so long? It is severely disruptive to my workflow.

There is even a pull request made by someone else that patches this problem.

OpenAI, please check and fix this.

Mine was running tests/debugging on a docker network setups with 5 containers and volumes, rebuilding the thing on each tests run , disk had about 0.5T of docker build cache…