Regression: resumed conversations are single-writer locked across windows and machines
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.4k
- PR merge metrics
- PR metrics pending
Description
Summary
Codex CLI 0.155.1 appears to have introduced a single-writer lock for resumed conversations. Opening the same saved conversation in another Codex window or on another machine now shows:
🔒 This conversation is open in another app
R to Retry
Close it there and press R to continue here.
Before the recent update, multiple windows/machines could resume the same conversation and stay synchronized: messages entered on one client appeared in the other in real time.
Reproduction
- Start or resume a saved conversation in Codex CLI on machine A.
- Open/resume that same conversation in a second Codex window or on machine B.
- The second client is blocked by the lock message above.
- Continuing requires reaching machine A and manually closing the particular Codex session/process.
Expected behavior
Support safe multi-client continuation using one of these approaches:
- serialize writes while broadcasting updates to all connected clients;
- provide a visible “Take over” / “Transfer conversation” action that atomically hands ownership to the new client;
- provide a remote “close/stop this conversation” action from the session list;
- at minimum, automatically recover from a disconnected/stale client lease.
A user should not need preconfigured SSH/tmux access to another machine, or process-level intervention, merely to continue their own conversation while away from home.
Impact
This is a major regression for cross-machine workflows. If the original machine is inaccessible, the user cannot continue the conversation from the second machine. The current recovery message says to close the conversation elsewhere, but does not provide a practical way to do that remotely.
Environment
- Codex CLI 0.155.1
- Linux
- ChatGPT-authenticated Codex session
- Regression from the behavior before the recent update
Please investigate the new conversation writer-lock behavior and restore safe real-time synchronization or add an explicit takeover/handoff mechanism.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the writer-lock behavior described for Codex CLI 0.155.1 on Linux with the same saved conversation open in two windows or machines. No file, test, or entry point is named; done would require restoring safe multi-client synchronization or implementing an explicit takeover, remote close, or stale-lease recovery path.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- cli, distributed-systems
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100