Mesa 7i96 LinuxCNC freeze /Init Led lights up after some time

More
15 Nov 2022 16:22 - 15 Nov 2022 16:26 #256779 by apfelschorle
Hello, after fixing my ethernet connection to the mesa board I'm facing a new problem. In the past I never get these problems with my linuxcnc configuration with my pc system. This is my normal working procedure. Starting the system which powers up my mesa board and everyting boots up correct and all led's behave as expected. After turning LinuxCNC on and running it for a while (no jogging, no jobs ect) linuxcnc and the hole linux system freezes. Mouse and keypad also dont work at all. Also the red led CR11 /Init turns on. The manual says that CR11 is also used to signal HostMot2 watchdog bite when LinuxCNC exits, but here it's freezing. The whole system freezes only when LinuxCNC is running for a while.

As mentioned in other threads here are the tmax values and the cpu clock frequency if there is something strange?
   
Kernel Version:
 

Also only one I saw this two messages on different days, but never again just the freeze
   

What I tried is to remove RAM sticks one by one to check if its comes from these but nothing, end of the week I get a ram stick from a friend and will try again. Also the cpu temperature looks normal round about 35°C. The 5V supply for the mesa board is also ok, tracked that no fluctuations also it is a high quality DIN rail power supply.

I hope someone has some ideas what I can do or check.
Attachments:
Last edit: 15 Nov 2022 16:26 by apfelschorle.

Please Log in or Create an account to join the conversation.

More
15 Nov 2022 16:40 #256781 by PCW
Yes, a red /INIT light is signalling a watchdog bite.

This is probably secondary to the HOST PC crashing, and
subsequent loss of communications. Simple losing
communications with the 7I96 will not cause the host PC
to crash.

The PC crashing sounds like a hardware issue.

Please Log in or Create an account to join the conversation.

More
15 Nov 2022 16:52 - 15 Nov 2022 16:59 #256782 by apfelschorle
Thanks for the response, I also thought this could only coming from a hardware issue, thats why I run the linux system for a while without running the linuxcnc app. I think it was half an hour or something also opened some programs and linux latency test was running and the system doesn't freeze in this time.

Maybe I have to do a ethernet stress test? do a ping loop? Do you have any idea how to do this or anything else to check? Or any way to use the linuxcnc live usb stick to test the connection between the mesa board and another pc without installing linux?
Last edit: 15 Nov 2022 16:59 by apfelschorle.

Please Log in or Create an account to join the conversation.

More
15 Nov 2022 17:19 - 15 Nov 2022 17:21 #256785 by tommylight
Since you say you checked the memory, check the processor for heat. Removing the CPU cooler and repasting it can sometimes fix this.
Next is the power supply.
Edit:
Also moved to "computers and OS's" since it is not a Mesa driver board issue.
Last edit: 15 Nov 2022 17:21 by tommylight. Reason: moved

Please Log in or Create an account to join the conversation.

More
16 Nov 2022 21:55 - 16 Nov 2022 21:56 #256903 by apfelschorle
So I had the time to setup linuxcnc with the live cd on a second pc.

Same results here, the /Init led turns red signaling the watchdog bite. Also I get the folowing errors. The difference is that I the whole system don't freeze like on the other pc system.Keyboard mouse and linuxcnc app works fine expect from that the cnc cant move, no jogging possible. The other pc freezes even when linuxcnc is not started ,so another pc problem.
 


Do you have more ideas what I should try?

PS: @tommylight can you move back to the old topic (Mesa driver board) because its a mesa driver board issue? So more people can help me which maybe had the same issue
Attachments:
Last edit: 16 Nov 2022 21:56 by apfelschorle.

Please Log in or Create an account to join the conversation.

More
16 Nov 2022 22:44 #256905 by tommylight

PS: @tommylight can you move back to the old topic (Mesa driver board) because its a mesa driver board issue? So more people can help me which maybe had the same issue

No, it is not a driver board issue, not yet. It is a PC issue, so:
-Disable hyperthreading in BIOS, disable power saving features, C states, etc.
-Test again.
You already have everything working and Mesa board communicating properly, just the latency is causing the link between them to fail.
If you need help with BIOS settings, just take some pictures and post them here.
The following user(s) said Thank You: apfelschorle

Please Log in or Create an account to join the conversation.

More
16 Nov 2022 23:03 #256909 by scotth
I'm seeing the problem with my 7i76E also. It is a Linux RT problem, not a Mesa problem. It is random and never seen a failure while running.
The following user(s) said Thank You: tommylight

Please Log in or Create an account to join the conversation.

More
17 Nov 2022 06:36 #256916 by apfelschorle
@tommylight
Ok, I did some screenshots but dont really find any options which covers the c state or other modes which should disabled? 
   

@scotth
How did you solved that? Do you use another kernel? 
Attachments:

Please Log in or Create an account to join the conversation.

More
17 Nov 2022 13:43 - 17 Nov 2022 13:44 #256929 by PCW
At the minimum, disable Intel Virtualization Technology, VT-d, Enhanced SpeedStep, and Turbo Mode.
Last edit: 17 Nov 2022 13:44 by PCW.
The following user(s) said Thank You: apfelschorle

Please Log in or Create an account to join the conversation.

Moderators: PCWjmelson
Time to create page: 0.347 seconds
Powered by Kunena Forum