[BUG] MCP Assistant (qmcp) causes nmd.pyc process leak → NAS freezes (no SSH/SMB/GUI access

Version: qmcp 0.10.0.797
QTS Version: QTS 5.2.9.3451
NAS Model: TS-463U


Symptom:
After starting qmcp, the NAS becomes completely unresponsive within 1-2 hours:

  • No SSH access
  • No SMB/network shares
  • No Web GUI
  • NAS still pingable
  • Hard power-off required (power button unresponsive)

Root Cause:
qmcp triggers repeated spawning of nmd.pyc (Network Manager Daemon) processes without terminating previous instances. The process count grows continuously until the system can no longer accept new connections.

Evidence:

$ ps aux | grep nmd.pyc | wc -l
15  ← growing continuously after qmcp start

After stopping qmcp and killing all nmd.pyc processes:

$ ps aux | grep nmd.pyc | wc -l
0  ← stable, no new processes spawned

TLS errors found in qmcp.log (likely trigger):

http: TLS handshake error from 127.0.0.1: remote error: tls: bad certificate
http: TLS handshake error from [::1]: local error: tls: bad record MAC
User not authenticated or username not found in context: error=no authentication found in request

Timeline:

  • qmcp starts at boot (~08:00)
  • nmd.pyc leak begins immediately
  • NAS freezes after ~1-2 hours
  • Reproduced twice on consecutive days

Workaround:
Keep qmcp disabled until fixed.

Request:
Please fix the TLS retry loop in qmcp that spawns nmd.pyc without cleanup.

Thanks for the report! I’ll have our internal team analyze and verify the issue.

Unfortunately, we were unable to reproduce the issue you mentioned and may need more information. Could you please share the AI model you are using, the task you are performing, and the prompt you are using? Thanks!