Desktop app crashes with "AppServerManager RPC is not connected" on every logout

What version of the Codex App are you using (From “About Codex” dialog)?

26.818.5345.0

What subscription do you have?

Pro

What platform is your computer?

Microsoft Windows NT 10.0.26200.0 x64

What issue are you seeing?

The desktop app crashes into a generic "Oops, an error has occurred" screen immediately after any logout event (manual or automatic). Also seeing a broader auth/performance problem that appears related.

Full error from the app's own logs, captured right after a logout:

[AppServerConnection] response_routed broadcastFallback=false conversationId=null durationMs=532 errorCode=null hadInternalHandler=false hadPending=true method=account/logout originWebcontentsId=1 queueWaitMs=0 requestId=e19be411-f9a0-413b-8062-ab7c6276f5e9 targetDestroyed=false

[electron-message-handler] error boundary errorMessage="AppServerManager RPC is not connected" errorName=Error errorStack="Error: AppServerManager RPC is not connected" name=App rendererWebContentsId=1 rendererWindowAppearance=primary rendererWindowFocused=true rendererWindowId=1 rendererWindowVisible=true

Separately, and possibly the root trigger: I'm seeing persistent 401s on GET /backend-api/accounts/:param/settings despite a valid token being present:

[electron-fetch-wrapper] desktop_fetch_auth_401 hadToken=true skipRetryReason=null target="GET https://chatgpt.com/backend-api/accounts/:param/settings"

This happens even immediately after a completely fresh sign-in following a full local session reset (cleared Login Data, Network, Partitions, Local Storage, Session Storage). Requests are also taking 24-55 seconds to complete, and specific older conversations fail to resume with:

[AppServerConnection] Failed to load older thread history conversationId=01a03064-f334-78a2-a143-0649307b256b errorMessage="no rollout found for..."

I'm filing the account/auth portion separately through OpenAI support since it likely needs server-side investigation, but including it here since it may be the actual root cause of the logout crash above.

What steps can reproduce the bug?

1. Launch the Codex desktop app while signed in.
2. Trigger a logout, either:

a. Manually via Settings → Log Out, or
b. Wait for it to happen automatically (in my case, triggered by repeated 401s on the account settings endpoint — see the linked issue description for detail)

1. The app immediately crashes to a generic error screen with a "Try again" button, within roughly 0.2-0.5 seconds of the logout request completing.

This is 100% reproducible on my machine via both paths (a) and (b) above, across multiple separate sessions today, including after a full local auth/session data wipe and fresh re-login.

Example session/conversation IDs from my logs showing the related slow-response and "no rollout found" symptoms (in case useful for server-side log correlation):

• conversationId=01a03064-f334-78a2-a143-0649307b256b (durationMs=24172, "no rollout found")
• conversationId=01a02f17-051a-7c83-971b-533dd12d81e9 (durationMs=54680)

Not applicable: token limit / context window usage — this isn't related to model context size, it's an auth/session-layer issue occurring before any model interaction.

Environment: Windows 11 Home, build 10.0.26200 (64-bit) — Microsoft Windows NT 10.0.26200.0 x64. App version: OpenAI.Codex_26.818.5345.0 (confirm against your About Codex dialog).

What is the expected behavior?

Logging out (whether manual or automatic) should return the user to a normal sign-in screen. It should not crash the renderer or show a generic "Oops, an error has occurred" page.

Separately, authenticated requests to /backend-api/accounts/:param/settings should succeed when a valid token is present, rather than returning 401. Conversation loads should complete in a reasonable time rather than 24-55 seconds, and previously-existing conversations should not return "no rollout found" errors.

Additional information

This is not an isolated one-off — it reproduced identically across many separate app sessions today, including:

• Before any local troubleshooting (original spontaneous crash)
• After a full local session/cache reset (cleared Login Data, Network, Partitions, Local Storage, Session Storage) followed by a completely fresh sign-in
• On both an automatic forced logout (triggered by the 401s) and a manual logout via Settings

The fact that it persists identically after a full local wipe + fresh login strongly suggests the root cause is server-side session/auth state tied to my account, not a corrupted local install. I've also opened a support ticket with OpenAI directly about the account-side symptoms (401s, slow responses, "no rollout found" errors), since that likely needs access to server-side logs this bug report can't provide. Happy to provide additional local log excerpts if useful — I have full session logs from every reproduction today.