A good way to debug this is to figure out which part is hanging: 1. browser/UI 2. Prefect API/server 3. database behind the API 4. reverse proxy / auth / network A minute-long “freeze” on reload usually suggests the page is waiting on an API request or the browser main thread is blocked. What I’d do first 1. Open browser DevTools and reload the page - Open Network tab - Check Preserve log - Reload the Deployments page - Look for requests that stay Pending for a long time or return errors In particular, watch for deployment-related API calls. In Prefect 3.x the UI will request deployment data from the API, so if one of those hangs for ~60s, that’s a big clue. What to note: - exact request URL - status code - total duration - response body, if any If requests are fast but the page is still frozen, that points more to a frontend/browser issue. 2. Check the browser console In the Console tab, look for: - JavaScript errors - repeated warnings - failed fetch/auth/CORS errors - out-of-memory or rendering errors If you see a stack trace on reload, that’s very useful. 3. Check Prefect server logs while reproducing it If you’re self-hosting Prefect Server, reproduce the issue while tailing logs. If you start the server manually, run it in the foreground and watch logs during reload. If it’s in Docker/Kubernetes/systemd, check container/service logs. You’re looking for: - slow requests - database timeout messages - worker timeouts - proxy/auth errors - 5xx responses 4. Test the API outside the UI Try calling the API directly for deployments and compare response time with the UI freeze. For example, if you can access the API docs or use curl against your Prefect API, test the deployments listing endpoint and see whether it also takes ~60 seconds. If direct API calls are also slow: - issue is likely API/database side If direct API calls are fast: - issue is more likely UI/browser/proxy related 5. Try another browser / incognito / disabled extensions This helps separate UI bugs from local browser state issues. - incognito/private window - different browser - disable extensions/ad blockers - hard refresh / clear site data If the problem disappears there, it may be cached assets or an extension conflict. 6. Compare old UI vs new UI You mentioned the old/current stable UI. If possible, try the newer UI as a comparison. If: - old UI freezes - new UI does not then this is likely a UI-specific issue rather than a backend issue. 7. Check whether the number of deployments is large If you have a lot of deployments, the page may be doing expensive filtering/counting/loading on refresh. This can expose: - slow database queries - missing DB maintenance - API performance regressions If self-hosted with Postgres, DB performance matters a lot here. 8. If self-hosted, inspect the database Useful checks: - CPU / memory pressure on DB host - slow query logs - blocked queries / locks - table bloat / need for

VACUUM ANALYZE

- overall deployment count and flow/deployment table size If the freeze lines up with slow deployment queries, that’s likely the root cause. Practical isolation matrix - Network request pending for ~60s → API, DB, proxy - Network request finishes quickly, but tab is unresponsive → frontend/browser - Only one browser affected → local browser/cache/extension - Old UI affected, new UI not affected → old UI bug/regression - API logs show timeout / slow SQL → backend/database issue Things worth collecting If you want to narrow this down quickly, gather: - Prefect deployment method: self-hosted server or Cloud? - browser + version - whether this happens in all browsers - approximate number of deployments - server logs during one reload - screenshot of DevTools Network waterfall - any Console errors Likely next steps / workarounds - try newer UI as a workaround - upgrade from 3.7.8 if possible, since UI/perf bugs may already be fixed in later 3.x releases

- if self-hosted, check DB/query performance first I couldn’t confirm a matching existing GitHub issue from tools because GitHub search was unavailable just now, so I don’t want to invent a known bug reference. If you want, I can help you debug this step by step. If you share: - whether you use Prefect Cloud or self-hosted Prefect Server - how you run it: Docker, Kubernetes, bare process - what you see in Network and Console on reload I can tell you exactly where the problem is most likely coming from.