CoreELEC flooding network with NTP lookups

Hi,

Why is CoreELEC behaving like this?
Is it by design or a bug?

I have entered my own NTP server (on my LAN) in CoreELEC settings but it does not respect that entry, obviously.
My firewall blocks any attempt to contact other NTP-servers.
Can I make CoreELEC to stop this?

CoreELEC: 21.1-Omega_nightly_20240723
Hardware: AM6b+ / Amlogic-ng.arm

No logs but here’s some screenshots from my PiHole:



Not enough information.
https://wiki.coreelec.org/coreelec:ce_support

KODI debug level doesn’t seem to show anything (to me).

Linux journalctl on the other hand sees the issue.
https://paste.kodi.tv/iciwimifuc.kodi

The issue usually stops when rebooting the device.

Registered purely to add a +1 to this.

(Almost) same issue, same device. Only difference is that I have a firewall rule in place to redirect all NTP traffic to my own server. Happens with all nightlies I have tried, including 21.1 stable. Seeing over 3.5 million queries to *.pool.ntp.org within a few days.

In my case, that’s 16 queries/second, every 5 seconds. About 12k/hour.

I am not able to reproduce, it ‘just happens’ at some point. A reboot fixes it.

+1,
CE21.1.1 asks the ntp servers every few seconds. I have a whole lot of ntp clients, and this is out of the ordinary that I have seen so far.

Interestingly CE still asks *.pool.ntp.org although you have provided 3 other ntp server addresses in the CoreElec config addon. I would expect it to just use the servers provided by the user. I also would expect it to use *.pool.ntp.org only if the user has not provided any other ntp servers at all.

If the user sets less than 3 other ntp server addresses, CE will ask *.pool.ntp.org for all address slots not set by the user.

If the user sets all 3 ntp server address slots, the repeated requests to *.pool.ntp.org stop after the device has fully booted up. During boot it still asks *.pool.ntp.org.

Do the deep analysis and send PR for fixing the issue.

I am not a developer. I am just a bit network-savvy.

In the meanwhile, a simple work-around is just filling all ntp server address slots, not leaving any empty.