在每个完成时被杀死的cgroup中运行每个编码代理会话
Run each coding agent session in a cgroup that is killed on completion
Codex任务完成。一位开发者构建了一个代理生成器，该生成器在每个编码代理会话中启动其自己的cgroup，并在任务完成后杀死剩余进程，还有一个清除陈旧tmp文件的作业.
做了什么
The developer traced heavy swapping to a RAM-backed tmpfs /tmp and to processes that earlier coding agents had forked and abandoned. They added a frequently run job that deletes tmp files unused for about two hours and not held open by any command. They also built their agent spawner to launch each session in its own cgroup, killing every process in it when the task finishes.
效果如何
The source gives no measurements. It says the cleanup job and per-session cgroups addressed the swap usage and the leftover processes, but it is a brief forum comment with no before-and-after figures.
值得注意
Check whether swap is really coming from a RAM-backed tmpfs /tmp before blaming the agents; the cleanup job only removes files with no open handles.
试试这个 Prompt
In my repo at [path], write a small agent spawner script for [Linux distro or systemd setup] that launches each coding agent session in its own cgroup and kills every leftover process in that cgroup when the task finishes or fails. Also add a scheduled job that deletes stale files older than [number] hours in [tmp directory]. Finish by running a test session that leaves a background process behind and showing me that it gets cleaned up. Before deleting any files or killing processes outside the test, show me what would be removed and wait for my approval.