[Bug] Backend is running but temporarily not responding on port 3900. Waiting for reco

Jokeryt27

<!-- Click Submit at the bottom of this page to file the issue. Review the auto-captured environment info below and add anything about what you were doing when the bug happened. --> ## Describe the bug <!-- e.g. "Synthesize failed in Design mode after picking Narrator personality" --> ## Error ``` Backend is running but temporarily not responding on port 3900. Waiting for recovery. ``` ## Environment **App:** VoiceStudio 0.5.6 **Shell:** Electron ## Backend reachability **Connection:** `managed local` **Last backend response:** 35 s before this report ## Current backend failure Stage: failed Managed: true Backend is running but temporarily not responding on port 3900. Waiting for recovery. ``` … (truncated) 127.0.0.1:53277 - "GET /model/loaded HTTP/1.1" 200 OK INFO: 127.0.0.1:50308 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:50308 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:50308 - "GET /model/loaded HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /dictation/prefs HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /workers/target HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/loaded HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/loaded HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /profiles HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /workers/target HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /model/status HTTP/1.1" 200 OK INFO: 127.0.0.1:50308 - "GET /workers/target?op=tts HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /system/notifications HTTP/1.1" 200 OK INFO: 127.0.0.1:53277 - "GET /batch/jobs?status=active&limit=100 HTTP/1.1" 200 OK ``` ## Recent actions ``` 10:50:52 view:home 11:00:08 generate:clone:start 11:09:53 generate:clone:error 11:10:00 generate:clone:start 11:11:16 generate:clone:cancel 11:13:32 generate:design:start ``` ## What I was doing <!-- step-by-step would help us reproduce -->

RainbowJaveline

Hi @Jokery27, I'd like to investigate this issue. If possible, could you please assign #2490 to me? I noticed that the backend appears to be running and responding to several requests successfully, but the application eventually reports that it is temporarily not responding on port 3900 and waits for recovery. I'd like to trace the backend health/reachability check and recovery logic to understand whether this is caused by the backend becoming unresponsive, a timeout/health-check issue, or the recovery mechanism incorrectly detecting the backend state. I'll first try to reproduce the issue locally and trace the relevant backend monitoring and port 3900 handling before proposing a fix. If you have any additional logs or steps that reliably trigger the issue, please share them. Thanks!

debpalash

Fixed in #2499: the health probe no longer imports torch or queries the GPU driver on every check, a busy backend is reported as recoverable instead of failed, and slow or no-GPU PCs get a longer startup budget. This is on `main` and ships in the next release. If you still see it afterwards, reopen with the log from Settings → Logs → Backend.