Linuxcnc floods me with errors..

Hi guys,
From time to time get some errors due to which I need to restart linuxcnc and redo all the homing routine.
Last time I got these errors...
I am using a Mesa 7i76e so I assume the board does the step generation.

This set of errors usually means that there is a latency problem with the host PC
What has happened is that you have a latency error so large (> say 800 usec assuming 1 KHz base thread) that:

1. You get a latency error message
2. the watchdog bites, disabling all I/O pins so you get Sserial errors since the connected device does not respond

I would run the latency test on the servo thread only for at least a few hours to test
the host latency and also do a ping test of the 7I76E

Also what value do you have set in the ini file for servo thread ?
You should leave that at 1000000 ( one milion )

If you're talking about SERVO_PERIOD than than the value is as you said.
SERVO_PERIOD = 1000000

Here is my config file if you think something else may improve this.
github.com/85vmh/LinuxCncConfigs/blob/ma...gs/SyilX7/SyilX7.ini

That looks all good to me.

Have you run a latency test?

I just did:
Servo Thread: 32000 ns
Base Thread: ~82000ns

Are these values so bad that they can cause those errors?

No, but you should run a latency test with no base thread:

latency-test 1ms 1ms
or
latency-histogram --nobase --sbinsize 1000

also a ping test should be done

Either

ping 10.10.10.10
or
ping 192.168.1.121

depending on how your 7I76E is set

While trying to make a video on the issue I have with the probe screen repetability, I just got:
hm2/hm2_7i76e.0: error finishing read! iter=649051

Pinging 10.10.10.10 for a few minutes got a reply every single time, no request timeout.

Running latency-test 1ms 1ms with ping running, glxgears running and using the browser while typing this post, resulted in 57128ns.

What is the latency of ping replies ? It is written at the end of each reply line.