Mesa Communication Failure

Hi,

I have setup using 1 x 7i80DB and 2 x 7i77 boards to run a big machine. It's been running perfectly for about a month now and then last week all of a sudden the Mesa boards lost communication with LinuxCNC while the machine was sitting idle. Linuxcnc did NOT throw the usual error about communication, it just sat there looking like it was working and allowing on screen input. The only indication of a fault is that any commanded move would result in an instant following error since the drives lost power when the boards shut down the outputs.

The blinking communication lights were NOT on on the Mesa boards and the boards themselves did what they were supposed to do by turning off all of the outputs. We rebooted the whole system (computer and machine) and communication was back up and running and we pulled the ethernet cable which made LinuxCNC come up with the communication error as it normally would.

Has anyone seen this before? Any suggestions on how to stop it from happening again? Should I not be relying on LinuxCNC to detect if communication is lost?

Thanks,

Rob

I have not seen or heard of a unreported failure of this type (I have about 3 years of machine uptime with hm2_eth)

What linuxcnc version is this? Does it have the Ethernet packet loss hardening?
(does it have the packet-read-timeout parameter)

If it happens again can you capture the kernel log (via dmesg)?

I'm running 2.7.7 which I thought had the packet-read-timeout parameter.

If we see it again I'll capture the kernel log... but if you have some instructions on how to do that that would be great.

Thanks!

I would run linuxcnc from the command line to see if there are any errors there

dmesg > logfile.txt

Will save the current kernel log to logfile.txt t

What PC hardware and OS are you using?

(Debian, Ubuntu, Mint, and kernel version etc)

Ok I'll try that.
We're running another job today so hopefully if there are errors it'll pop up again and we can capture the logfile.

I'm running an older AMD Phenom x4 quad core CPU, low end discreet graphics (Nvidia I think) and onboard Ethernet for the connection to the machine and with Wifi for internet. I have Debian Wheezy installed for the OS since that seemed to be the only one that runs the Preempt RT kernel needed for the HM2_eth stuff. Maybe I'm way off on that...

You can can build newer Preempt-RT kernels if needed for new hardware/ newer OSes
(I'm running 4.1 through 4.6 kernels on Wheezy, Ubuntu 14.04, and Mint 17/18 here)

Do you know what Ethernet chip is used on the Motherboard?

My long term testing is with Intel and Realtek, I've had bad experience with Atheros and Broadcom