codex - 💡(How to fix) Fix Remote compact task consistently times out on CLI (author login), while API-compatible provider path works [2 comments, 2 participants]

What version of Codex CLI is running?

codex-cli 0.125.0

What subscription do you have?

ChatGPT (author login)

Which model were you using?

gpt-5.4，gpt-5.5

What platform is your computer?

Ubuntu Linux 22.04.01

What terminal emulator and version are you using (if applicable)?

SSH terminal (xterm)

What issue are you seeing?

Error running remote compact task: timeout waiting for child process to exit

After this error, the session becomes unusable and cannot continue.

What steps can reproduce the bug?

1. Use Codex CLI with author (ChatGPT) login

2. Start a long-running task (e.g., multi-file code analysis)

3. Let the context approach the token limit

4. Trigger:

   • Automatic compaction, or
   • Manual /compact

Result:

Compaction fails 100% of the time and the session becomes unusable

What is the expected behavior?

Compaction triggers a remote task
→ the task does not return in time (or stalls)
→ CLI throws:
   "timeout waiting for child process to exit"
→ session is effectively broken

Additional information

✔ Network connectivity is normal

Tested:

curl https://chatgpt.com/backend-api/codex/responses/compact

Returns quickly (HTTP 405 without proper method/auth), indicating connectivity is fine.

✔ Local environment is normal

• Node / npm versions are valid
• Fresh reinstall does not resolve the issue
• No unusual environment configuration

✔ Key observation: behavior differs by backend path

❌ Author login (ChatGPT backend)

Compaction → timeout → session unusable

✅ OpenAI-compatible API provider (via proxy / API key)

Compaction → "Context compacted"
→ session continues normally

This indicates that:

Different backend paths for compaction behave differently

✔ Cross-platform difference

• Ubuntu codex-cli → consistently fails
• Windows Codex (same account) → does not crash (though quality may degrade)

This suggests inconsistent error handling or fallback behavior across clients.

Impact

- Long-running sessions are effectively unusable in CLI with author login
- Compaction failure permanently breaks the session
- Blocks real-world engineering workflows

Questions

• What exactly is the "remote compact task"?
• Why does it fail consistently on the ChatGPT backend path?
• Is there a timeout / retry / fallback mechanism?
• Why does failure result in a fatal session state instead of recovery?

Workaround

Using an OpenAI API key via an API-compatible provider (proxy) avoids the issue

Summary

Compaction via author login backend consistently times out and breaks sessions,
while API-compatible provider paths complete successfully.

This strongly suggests a backend-specific compaction issue rather than a local environment problem.