VMs no longer working

📝 Bug Report Draft: Missing novnc_proxy / VM Console Fails on 7.1.4

System:

• Unraid OS 7.1.4 (upgraded in place from 7.0.x)

• Hardware: [insert your board/CPU/RAM if you want, not strictly required]

• Not running in Safe Mode

Summary:
On one of my Unraid 7.1.4 servers, all VM Console sessions fail with “Failed to connect to server”. This appears to be due to the novnc_proxy script being missing from the Dynamix VM Manager plugin folder. The same version (7.1.4) on a different server does not exhibit the problem — VM consoles work fine there.

Steps to reproduce:

1. Start any VM (tested with Windows 11 and Debian 12).

2. Click Console from the VMs tab.

3. Browser attempts to connect to /wsproxy/5700/ or /wsproxy/5701/.

4. Console fails with “Failed to connect to server”.

What we expected:
noVNC proxy should launch on port 570x, bridge to QEMU VNC (590x), and display the VM console in browser.

What actually happens:

• QEMU VNC listener appears correctly on 127.0.0.1:5900/5901:

  ss -lntp | grep 590
  LISTEN 127.0.0.1:5900 qemu-system-x86
  

• No novnc_proxy process ever spawns:

  pgrep -af novnc_proxy   # returns nothing

• No websocket listener on 570x:

  ss -lntp | grep 570# only unrelated docker-proxy entries

• nginx logs show connection reset when proxying to 127.0.0.1:570x.

Diagnostics / further checks:

• Verified Safe Mode is not enabled.

• All plugins (Community Apps, etc.) load normally.

• Verified the novnc proxy script is missing:

  ls -l /usr/local/emhttp/plugins/dynamix.vm.manager/scripts/novnc_proxy
  ls: cannot access '.../novnc_proxy': No such file or directory
  

• Rebooted multiple times → issue persists.

• Compared against another 7.1.4 server → novnc_proxy exists there and consoles work normally.

• Attempted workaround by exposing direct VNC in VM XML (listen=0.0.0.0 port=5999) → works from an external VNC client, confirming the VMs themselves are healthy.

• Tried to reinstall VM Manager via Plugins tab, but VM Manager does not appear as a standalone plugin (expected, since it’s part of the core Dynamix suite).

• Tried to force reinstall 7.1.4 via Tools → Update OS, but Update reports “You already have the latest version” and offers no reinstall option.

Impact:
On this system, browser-based VM consoles are completely unavailable. All VMs must be accessed via external VNC/SPICE clients, which is not normal behavior.

Suspected cause:
Corruption or missing files from the 7.1.4 upgrade on this system (at least /usr/local/emhttp/plugins/dynamix.vm.manager/scripts/novnc_proxy). The rest of the Dynamix VM Manager appears to function (VMs start/stop/edit normally).

Requested resolution:

• Provide a way to re-install the Dynamix VM Manager / noVNC component without reinstalling the entire OS from USB.

• Or confirm if this is a packaging bug in 7.1.4 that can cause incomplete deployment of novnc_proxy on some systems.

• Ideally, Update OS should offer a “Reinstall Current Version” option to recover missing core files.

Workarounds used so far:

• Direct VNC via modified <graphics> XML (listen=0.0.0.0 port=5999) → confirms VMs are healthy.

• SSH tunnel to 127.0.0.1:5900 works as well.

• None of these restore browser console, only external access.

s3v3nd34dly51ns-diagnostics-20250904_1522.zip