Spindle started during tool length probing

Hello,

sorry for asking without being able to give any valuable details. I have no logs, and the issue I had is not reproducible. But it was quite disturbing, so I'll try my luck and ask if this maybe rings a bell to someone out there.

I have installed LinuxCNC 2.9.2 on a Raspberry Pi 4, the image from the LinuxCNC page. Before I had 2.8 installed, which flawlessly cut many chips.

I'm running (as before) Gmoccypy.

After installing, I took over my config files from 2.8 and was surprised that it actually worked quite nicely out of the box. I installed ProbeScreen for LinuxCNC 2.9 (github.com/verser-git/probe_screen_v2.9); before I already had installed ProbeScreen for LinuxCNC 2.8).

Already cut some chips including tool length sensing etc, everything just flawless.

But today something VERY strange and disturbing happened: LinuxCNC had been running for some days, machine was homed, so I just loaded a GCode file (but did not start it yet) and then wanted to first touch off the tool probe with my 3d probe (which I always do and had done several times already with the new set up).
It drove above the tool probe, approached, touched, retracted, touched slowly, retracted - and then started the spindle motor!! Of course the probe cable was ripped apart and I was left shocked.

Then I tried several times again (with the 3d probe not attached and simulating touch-off by hand). It did the very same each time. Then I restarted LinuxCNC - and now it behaves as usual again.

Normally, when such things happen, I just frown a bit and am glad that the issue resolved. But here I am really curios how this could actually happen.

What I've seen is that I have repeated errors on the MB2HAL, I think every 2nd transaction does not work. But even with these error, the spindle works flawlessly (I will take care on these of course). My first guess was indeed that there was some glitch on the modbus, causing the spindle to start, but as it started after each probing repeatedly I don't think this is the issue.

I think, if I don't explicitly collect logging data from stdout, I don't have any logs, right? I should have stored the stdout (I even had LinuxCNC running from command line!)

Any idea anyone how this could possibly happen? Again, sorry for the sparse information, I know it's hard to tell anything without seeing any logs.

BR Simon

It may be related to this:

github.com/LinuxCNC/linuxcnc/issues/2764

If you have unexplained behavior again, one way to
save at least the hal state is to run:

halcmd show all > log.txt 

in a terminal