codex - 💡(How to fix) Fix Desktop should update or restart remote Codex app-server when version is too old [2 comments, 2 participants]

Codex on this environment is out of date. Update to 0.128.0 or newer. Current version: 0.126.0-alpha.15

ssh <remote> 'codex --update && codex --version'\n# then restart/kill the stale remote codex app-server process so Desktop reconnects to the new binary\n```\n\nThat is hard to discover, and the user often loses access to active remote sessions until they manually SSH in and know which process/cache to update.\n\n### Expected behavior\n\nCodex Desktop should do one of the following when it detects a remote version mismatch:\n\n1. Show an explicit action such as `Update remote Codex and restart app-server`.\n2. Run the remote update/restart automatically when the remote host is reachable and the user has permission.\n3. If automatic update is intentionally unsafe, provide copy-pasteable commands that update the remote CLI and restart the specific app-server process Desktop is using.\n\nThe flow should distinguish these states clearly:\n\n- installed remote `codex` is too old\n- installed remote `codex` is new enough, but the running app-server process is stale\n- Desktop itself is too old\n- update failed because the remote updater failed\n\n### Why this matters\n\nThis has shown up for remote app users after Desktop updates raised the minimum app-server version. The remote pod/host can be healthy and SSH-reachable, but Desktop becomes unusable until the user manually updates the remote CLI cache and restarts the old app-server.\n\n### Related\n\n- #19300 covers stale remote app-server reuse and diagnostics.\n- This request is specifically for a product-supported update/restart path when Desktop already knows the remote app-server version is below the minimum.