firefox restart the page on an endless loop

[michel-io]

Prerequisites

• I am running the latest version of ComfyUI
• I have searched existing issues to make sure this isn't a duplicate
• I have tested with all custom nodes disabled (see how)

What happened?

i recently updated the front end of my comfyui to the lastest version, the server loads well, but when the page starts to load, then reboots on an endless cicle.

last known good version it's 1.30.6

the error happened on a firefox 145.0.2 running on windows 10

Steps to Reproduce

How is this affecting you?

Crashes ComfyUI completely

ComfyUI Frontend Version

1.32.9

Browser

Firefox

Console Errors

Logs

Additional Context

No response

┆Issue is synchronized with this Notion page by Unito

[dosubot]

Hi @michel-io! I'm Dosu and I’m helping the ComfyUI_frontend team.

This endless reload loop in ComfyUI Frontend 1.32.9 on Firefox is likely caused by recent changes to how the frontend handles asset loading errors and authentication/session issues. Specifically, the frontend now listens for Vite's vite:preloadError event and triggers a full page reload if assets fail to load or if authentication fails. If these errors keep happening—due to browser-specific asset loading quirks, frontend-backend version mismatches, or persistent 401 errors—the page can get stuck in a reload loop, especially in Firefox, which may handle these events differently than Chrome or Edge [source].

There are reports of similar issues in recent versions, and fixes for infinite reload loops were merged after 1.32.9, so this is likely a regression in your current version [source].

Troubleshooting steps:

• Fully disable all browser extensions (not just for localhost), as privacy/user-agent switchers and other extensions have caused similar issues even when "disabled" for local sites.
• Clear your browser cache and cookies.
• Check the browser console and network logs for asset loading errors or repeated authentication failures.
• Make sure your backend and frontend versions are compatible.
• If possible, try a newer frontend version (1.34.0 or later) where similar reload loop bugs have been fixed for some users [source].
• If you need to keep working, temporarily reverting to 1.30.6 should avoid the regression.

If the problem persists, capturing and sharing any console or network errors will help maintainers diagnose the root cause.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

  

[michel-io]

i did all the things recommended, still with the same problem, i didn't post logs because i don't see anything out of place, but still the problem persist, getting a browser log to share is difficult, because the reboot loop don't let me make screen captures

[dosubot]

You've already done all the standard troubleshooting, so at this point the endless reload loop in ComfyUI Frontend 1.32.9 on Firefox is almost certainly a regression triggered by persistent asset loading or authentication/session errors. This behavior is baked into the frontend: if asset loading fails or the backend returns repeated 401 errors, the app reloads the page, and if the error keeps happening, it gets stuck in a loop—Firefox seems especially sensitive to this [App.vue, GraphCanvas.vue].

Recent fixes for these reload loops were merged after 1.32.9, so upgrading to 1.34.0 or later may resolve it for you [PR #6410, issue #6959]. If you can't upgrade, there isn't a built-in way to disable the reload-on-error logic in 1.32.9.

If you're running ComfyUI behind a subpath or reverse proxy, there's a known routing break starting with 1.32.2 that can cause endless reloads or blank screens in Firefox. Fixing this requires rebuilding the frontend with the correct base path using Vite's --base flag or the BASE_URL environment variable [issue #6995].

For deeper diagnostics, try capturing a HAR file or enabling verbose logging in Firefox's dev tools to check for repeated 401s or asset failures. If you can share those details, I can help interpret them and narrow down the root cause.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

  

[michel-io]

my rig runs locally, and the front-end it's the latest. I install with pip install comfyui-frontend-package --upgrade

off all the logging profiles, what should i use? i tried the networking log and it's massive, like 125mb of logs just for a couple of loops, and in the end i didn't found nothing there

[dosubot]

For this kind of reload loop, the most useful info comes from the browser Console tab, not the Network log. Focus on the Console and filter for errors and warnings—look for asset loading failures, 401/403 authentication errors, or module specifier errors. In Firefox, you can click the filter icons at the top of the Console to show only errors and warnings, which keeps the log small and relevant.

If the reload loop makes it hard to capture logs, try copying the first few error lines quickly or use a screen recording. If nothing appears in the Console, that's also useful to know and may point to a browser-specific or regression bug.

There’s no special debug mode or logging profile for the frontend itself—backend/server logs can be made more verbose via the 'log-level' setting, but that won’t affect what you see in the browser Console. If you can grab any Console errors or warnings, I can help interpret them and narrow down the cause.

Reference: Console errors are the key clues for diagnosing reload loops, not Network logs

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

  

[michel-io]

this it's the best that i can do, the loop don't give much time

[dosubot]

The endless reload loop with the stuck spinner and repeated JS requests is caused by the frontend's error handling: if asset loading fails or authentication errors (like repeated 401s) occur, the app silently reloads the page before any error appears in the console. This is why you see the spinner and network activity but no clear error message—the reload happens too early for logs to be captured, especially in Firefox, which can trigger these reloads more aggressively than other browsers [App.vue, GraphCanvas.vue].

If you can't capture console errors, try a screen recording or remote debugging (e.g., open DevTools on another device). Testing in Chrome or Edge can help confirm if this is Firefox-specific. If you’re running behind a subpath or reverse proxy, there’s a known routing break in 1.32.9+ that causes this exact symptom in Firefox—fixing it requires rebuilding the frontend with the correct base path using Vite’s --base flag or the BASE_URL environment variable [issue #6995].

If none of these apply, upgrading to 1.34.0 or later may resolve the issue, as fixes for reload loops and blank screens have landed in newer versions [issue #6959]. If you’re stuck, reverting to your last known good version is a safe fallback.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

  

[michel-io]

filtering to xhr

[michel-io]

the last entry goes 404 and the loop happpends

[dosubot]

A 404 on the /userdata endpoint by itself shouldn't trigger a reload loop—the frontend is designed to handle missing user data files gracefully and either ignore them or show an empty state. However, the 403 errors you're seeing on the "component" and "version" endpoints are more likely to be the real cause. Persistent 403s (Forbidden) on required endpoints can break frontend initialization and may trigger the reload logic, especially if they're related to authentication or a version mismatch between frontend and backend.

I'd recommend checking your backend logs for errors on those endpoints, making sure your backend and frontend versions are compatible, and reinstalling both if needed to rule out a corrupted install. If you haven't already, try clearing your browser cache and testing in another browser to rule out browser-specific issues. If the problem persists, the 403s are the key clue—those endpoints need to return 200 for the frontend to load properly.

References:

• Frontend code: 404s on /userdata are handled, not supposed to reload
• Reload logic is only triggered by asset preload errors or authentication failures, not 404s on /userdata

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

  

[michel-io]

i just test loading the page on opera and chromium an the problem happens there too

on the backend this log happens at every loop:

Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle: <Handle _ProactorBasePipeTransport._call_connection_lost(None)>
Traceback (most recent call last):
File "C:\Python312\Lib\asyncio\events.py", line 88, in _run
self._context.run(self._callback, *self._args)
File "C:\Python312\Lib\asyncio\proactor_events.py", line 165, in _call_connection_lost
self._sock.shutdown(socket.SHUT_RDWR)
ConnectionResetError: [WinError 10054] Se ha forzado la interrupción de una conexión existente por el host remoto

[jovan2009]

I'm not sure, but your problem it's possible to be similar to this #6959
Make sure you are running the latest git version of ComfyUI, my issue was solved on the server side, I think.