FTL Severe and Constant DNS Slowdown/Unresponsive in Proxmox LXC (v6.3.3) - Linked to Insufficient CAP_SYS_NICE permissions

[RtoRoc]

Hello,

I'm reporting a severe and constant performance issue where the DNS resolution provided by FTL becomes unresponsive for several seconds, leading to failed or extremely slow page loads on client devices.

Versions

• Pi-hole: v6.2.2
• Web: v6.3
• FTL: v6.3.3

Platform

• OS and version: Debian 12 (Bookworm)
• Platform: Proxmox LXC container (Unprivileged)

Expected behavior

The FTL DNS service should respond to queries quickly (milliseconds), regardless of internal background processes (like database maintenance or garbage collection).

Actual behavior / bug

The DNS resolution provided by FTL exhibits a severe and constant performance degradation (delays of several seconds) on all client devices using Pi-hole.

• Client browsing is consistently slow and often results in DNS timeouts or failed page loads.
• This performance issue persists continuously under the current setup.
• When clients bypass Pi-hole and use an external DNS server (e.g., 1.1.1.1), the navigation immediately returns to normal speed, confirming the issue is with FTL.

The root cause is strongly suspected to be the continuous failure to set process priority, as seen in the persistent log warning:
WARNING: Insufficient permissions to set process priority to -10 (CAP_SYS_NICE required), process priority remains at 0

This inability to set the priority means the critical DNS resolution thread is constantly competing with lower-priority maintenance threads (like the database garbage collector), causing continuous blocking and slow response times.

Steps to reproduce

1. Have Pi-hole installed in an Unprivileged Proxmox LXC container.
2. Set Pi-hole as the primary DNS for a client.
3. Browse the web.

The problem is constantly present under these conditions.

Debug Token

• URL: https://tricorder.pi-hole.net/KpCMI3zp/

Additional context

The low load average (Load: 1.29 / 1.00 / 0.89 as per the attached screenshot) confirms the issue is not resource exhaustion, but rather a scheduling/priority contention bug within FTL due to the lack of the CAP_SYS_NICE capability in the LXC environment.

[yubiuser]

Unprivileged Proxmox LXC container.

The unprivileged container seem not to be allowed to use SYS_NICE.
Have you tried to run it as privileged container or grant the specific permission?