openai / openai/codex

ChatGPT Sites publishing repeatedly disconnects before approval completes

Open
#40,753 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug codex-web connectivity
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

Description

Publishing an existing ChatGPT Site repeatedly fails before deployment begins.

The affected Site is:

Site slug: rankforge-leaderboard
Site name: RankForge Tournament Rankings
Production URL: https://rankforge-leaderboard.xxblad3.chatgpt.site/
Visibility: Public
I am the Site owner
Error

Every publishing attempt fails with an error similar to:

Network request disconnected after 1025 ms, before approval could complete

The duration changes between attempts. Other observed values include approximately 115 ms, 180 ms, and 241 ms.

Steps to reproduce
Open the existing rankforge-leaderboard Site project.
Make a source-code change.
Successfully run the Site in preview.
Request publication to the existing production Site.
Explicitly approve the deployment.
Publishing disconnects before deployment begins.
Expected behavior

The approved checkpoint should deploy to the existing Site while preserving its D1 database, environment variables, hosted secrets, and existing production URL.

Actual behavior

The publishing request disconnects before approval completes. The production Site remains on its previous version.

Troubleshooting already completed
Confirmed the Site exists and is active.
Confirmed I own the Site.
Confirmed the Site is public.
Confirmed its environment variables are present.
Confirmed hosted secrets remain configured and protected.
Confirmed the agent preview works.
Confirmed all 75 project tests pass.
Retried publishing multiple times.
Explicitly approved the deployment.
Retried using the equivalent of --user-approval-provided.
Updated the Sites workflow/plugin from version 0.1.22 to 0.1.29.
Retried from later conversations.
Submitted feedback through ChatGPT, but the feedback interface did not provide a session-sharing option or a session ID.

The failure remains consistent and occurs before any production mutation.

Impact

I cannot publish completed changes to an existing production Site. Preview works, but the production deployment cannot start.

Additional information

The prepared update includes archived-season scoring overrides and a new hidden All-Star qualification tracker. The failure does not appear related to application tests or runtime behavior because preview succeeds and the complete test suite passes.

The existing production version has not been overwritten or damaged.

I can provide screenshots or additional diagnostic information if needed. I have intentionally omitted API keys, secret values, database contents, and other sensitive information.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing publication from the existing rankforge-leaderboard Site project: make a source change, run the preview, request publication, and approve deployment. Compare the approval request with the reported disconnect timings and confirm that a successful result deploys the approved checkpoint while preserving the production URL, D1 database, environment variables, and hosted secrets.

Written by the indexing model from the issue text.

Assessment

Domain
cloud, release
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.