High CPU from node processes
Resolved 💬 3 comments Opened Mar 2, 2026 by aehlke Closed Mar 2, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex CLI is running?
0.106.0
What subscription do you have?
Pro
Which model were you using?
gpt-5.3-codex
What platform is your computer?
macOS Tahoe
What terminal emulator and version are you using (if applicable)?
iTerm2
What issue are you seeing?
I run several Codex CLI agents and have noticed that they often end up spawning many node processes that consume ALL available CPU, even long after the agents have stopped doing any work.
I am not running any other node processes. When I quit all agents, the node processes all go away.
<img width="828" height="574" alt="Image" src="https://github.com/user-attachments/assets/12bfbd3b-fe59-4148-985a-81399bb4e8aa" />
What steps can reproduce the bug?
I haven't figured out a pattern yet but I have been using the background terminal feature quite a lot. I also use plan mode. I do not use subagents.
What is the expected behavior?
_No response_
Additional information
_No response_
3 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
<img width="369" height="176" alt="Image" src="https://github.com/user-attachments/assets/6c6b0cf5-a061-4a8b-b426-7baa513fd6e0" />
the problem is still here:
issue: 9feda143-637f-40fc-a902-344f5e7d511c
Codex version: Version 26.311.21342 (993)
Device: MacBook Pro M1 Pro 16gb, macOS Tahoe 26.3.1 (25D2128)
I filed a narrower follow-up with sanitized local process evidence: #28794. The hot process in my latest case was the native Codex CLI worker itself, not just MCP helper children: no TTY, revoked stdio, stale after task_complete, about 831 CPU minutes before cleanup.