Whole GUI hangs up

Hello.

Last October we've updated LinuxCNC to 2.7.11. Firstly it worked like a charm.
But after about a week host machine started to hang up at the random time. Sometimes after couple minutes of work, sometimes after hours. Strange this was that the mill still worked as expected, carefully following the program. But GUI was just freezed. I couldn't even open linux console with Ctrl+Shift+F1 (or F2 ...). It looks like some user space issue with kernel, but I'm not sure.
Later I've raised latency to 30000. It looked too much but... it worked. Computer didn't hang up so I thought I managed that problem.
And last week same issue has returned.

Maybe someone had something like that? What could I miss? Where should I look? Can this thing happen due to the age of computer? Maybe HDD or other device problems?

Our milling machine is PFE 1500-PX.

Thank you.

Thank you for response.

No I haven't try SMI patch. I'll check what chipset our computer has. Maybe I should try it. Why is that bad idea? Can it destroy chipset? Is this a page I should follow? wiki.linuxcnc.org/cgi-bin/wiki.pl?FixingSMIIssues

This problem happens almost at every program. But amount of time before it stalls is pretty random. On the same job it can freeze at the different time. I've also tried find something useful in dmesg or syslog and had no luck.

By the way we have similar computer (they both had been bought about the same time) which couple days ago showed same behaviour. We use it pretty rare so it had no opportunity to show this. Until now.

Strange thing is that on the previous version (unfortunately I can't name it right now. If it can help, I can find it out) we had nothing similar to this. All started when I've upgraded LinuxCNC to 2.7.11.

I have a similar problem. I'd like to know if it's computer relater or not. I can make the gui hang during MDI mode when I issue a spindle rpm number and then select a x,y move from the history. If the spindle is not at the commanded speed it hangs if I try to re select the x,y move or select a different position. The whole MDI tab becomes blank (the list of historical m and g codes) and gui become non responsive. Keyboard keys also fails. The good thing is, no movement is executed. If I ramp the speed up manually to where I'm in range of spindle at speed the spindle at speed led on my panel is still indicating not at speed and no movement occurs. This is on a manual rpm control, vari-drive bridgeport mill with encoder feedback. I have been able to repeat this 3 times with similar results. Usually gui comes back to life after 30 seconds to a min or so. This has not occurred during execution of g code program in automatic mode or when I'm just jogging in MDI mode with spindle running at random speed. Am I confusing the system with my manual ramp up of the rpm?
This is on axis 2.8.0-pre1-2762-g42ff261 with mesa hardware. Attached is the dmesg file.

[ 37.073041] hm2_5i25.0: initialized AnyIO board at 0000:03:00.0
[ 37.164477] NMI: PCI system error (SERR) for reason a1 on CPU 0.
[ 37.164481] Dazed and confused, but trying to continue

Thank you,
OT

That's a hardware error, either a bad PC, bad 5I25, dirty PCI connector/5I25 contact fingers or bad ATX power supply

The 5I25 should only assert /SERR if it detects a parity error on the PCI bus

Thanks for the quick reply PCW. I'll pull the card and check contacts for starters and voltage. On debian, is there a way to monitor system voltage or get additional error messages to see if its PC related?

I believe I have:

spindle-at-speed => motion.spindle-at-speed

Yes, the pausing is normal behavior which works nicely during program execution. I have a contactor for now that enables the spindle and I usually get motion after spindle at speed becomes true.

I did some configuring today, machine was powered on for 3 hrs or so, launched linuxcnc numerous times and did another dmesg at the end and no errors came up. I did not dig into the pc or try to bring up the frozen screen. Is this type of parity error something that would be intermittent?