I need an answer

I am trying to finish up the construction of a CNC table using LinuxCNC, and I keep getting an error.  History this is a Dell SSF I3 2130 PC running 3.4 Ghz with 8 Gigs ram standard hard drive.  Loaded with Debian 11 Gnome Axis using a Mesa 7I76E.  I have disabled the Audio, C state, Hyper-thread, and sleep mode in the BIOS.  I have replaced the wireless keyboard and mouse with wired ones.  I have run Latency with 5 Glxgears running for about 10 minute and recorded and entered the value of 1006385 into Pncconf .  I get this error in random intervals, when homing, running a G-code or nothing at all. 
I have built other table with other software and this has cost me way more time then any other, I am sick and need help.
Here is a link for all the axis files from the disc.
drive.google.com/drive/folders/1HwH4B1Q4...G65iO?usp=share_link
 

A watchdog bite indicates issue with real-time

I think the default for a pncconf created configuration is 5 ms, meaning
that communications need to have been paused for at least 5 ms to get that error

This can be for a number of causes. The first thing I would try
is simply pinging the 7I76E card to see what the echo times are
 

Network latency is the most likely cause. Its different to the latency measured with latency-test

go buy a iron ferrite core wrap for you network cable , try that. could be close coupled RFI modulating the wire in a unkind way

www.amazon.com/uxcell-Ferrite-Cores-Supp...Filter/dp/B07YJWGFKZ

smc.collins wrote:

go buy a iron ferrite core wrap for you network cable , try that. could be close coupled RFI modulating the wire in a unkind way

www.amazon.com/uxcell-Ferrite-Cores-Supp...Filter/dp/B07YJWGFKZ

No, definitely debian kernel related from 5.10 to 6.1 as there are many data points from many people confirming this.

I looked over the code for the 7i76e, and the network drivers in general, there's a lot of places for overflows and memory problems. the newer _s versions of functions are not in use, and are only available in newer versions of C/C++. You may very well be correct. However, if the packets are getting noised up in traffic, this could be exaggerating or triggering the issue. If this is a known issue in newer Kernel revisions, then project should revert back to earlier kernel version in mean time as a update.

I did check the Ping and the time is about .2 to.3 msec for a return. I am not 100% sure but it seems the NIC card is a RTL8168 in the PC I bought.
As to the above post this problem can show up with no function activity, meaning the system just sitting idle. I have run simple G-codes a few time in a row and leave the system idle and come back to see the errors on the screen. But it also has failed at the start up and doing the homing routine. Just at random times.  This table is in my hobby shop with no other noise making equipment involved in the area.
I have looked for a ferrite core which I have a few and could not find on at this time, I will continue to look. (: (:
Thanks for the replies

Most evidence point to the RTK driver issue rodw mentioned.
Fixes are using the driver patches and a 6.x kernel
or reverting to 4.x real time kernel (if possible)

If you follow some of the steps in Rod's guide it's easy to see what NIC you have and what driver is actually loaded.

forum.linuxcnc.org/9-installing-linuxcnc...dware-eg-realtek-nic

Even works for a linux newbie.

sudo apt-get install ethtool  - gets you the ethernet tool to check the driver/firmware
ip a - gets you the NIC card identification number (or whatever it's name is)

once you have the name of the NIC card...

sudo ethtool -i enp2s0 (or "eth0" or whatever the NIC name is)

That'll tell you for sure what you've got.  Really sounds like another case of the realtek network issue.  I've abandoned attempts to get two different PC's working with LCNC because of it.