V_3.7.9 is free from this bug can confirm
Here’s a VIDEO
Now will be updating versions until i get the error again.
6 Likes
@CesarH Thanks for doing all this testing!!!
1 Like
Had the time and some spare esp32s so i dont mess my configs.
Now: its safe to update till version 3.7.17
V3.7.18+ are ones with the errors
This was with 3.7.18 flashing
As soon as i hit this tab, it halts the current job. And hardlocks the board (you need to powercycle it )
You endup with this error in the end
3 Likes
A few items to note in FluidNC changes…
This is the version where Mitch updated the Expressif libraries from 2.0.15 to 3.2, along with the Wifi and Webscoket code:
There were also multiple memory improvements in this version
Since the bug appears to involve issues around interrupts, maybe this commit might also be of note:
3 Likes
Going to be testing 3.7.17 next week for extended use.
My setup is as i said: AP mode with a pi4/touchscreen as a Fullscreen Kiosk. Rs485 controlled spindle (with the bart 6pack)
The jackpot runs the router with a relay. Going to make the time to try a 3d carve and see if i get some errors that way.
I highly recommend not messing with the config tab while running a job if you are running the newer version
All my testing was using V2 WebUI v3.7.12+ was using the updated webui as linked in the release
3 Likes
Excellent testing! Thank you.
Thanks for the insight. Are you pushing these observations back to the FluidNC team?
1 Like
I may try to formulate a message to them later, but I’m currently out of town on business, so I’m not sure how much time I’ll have to track down the right avenue to push it to them.
3 Likes
That would be a bummer if it was the library update, I really like things to be relatively up-to-date.
I guess…stability IS far more important.
2 Likes
Easiest path would be to remove the ISR to IRAM commit and rebuild 3.8.3 locally and test again.
If that doesn’t do it , then just start with 3.7.17 and add individual commits until it reproduces…
but I’m not home for a few weeks to try it…
2 Likes
If someone else wants to try, building FluidNC is pretty easy.
I can offer spotty remote VSCode and Git support to help…sometimes 
3 Likes
Ryan, I cannot find where I asked, but you had told me what the safe version was and a link. Should you maybe make sure that link points at 3.7.17?? For now
just found my post, I updated to .18. I did have problems with it the last time I used it. (But I was mistaken, there was no link).
I can try this later.
3 Likes
so interesting you guys mention Iram, found this in the discord:
So what are they talking about removing spindle story from Iram?
1 Like
Awesome.
It’d be great if it’s just the ISR commit, as that might be the easiest to convince to rollback…
Otherwise, the only thing that I noticed at a glance is that these 2
will have to be rolled back in together if you have to resort to starting at 3.7.17 and moving forward from there. Otherwise I’d maybe suggest trying the git bisect method.
I’ll probably hold off on sending any information to Mitch until we get a better answer if someone is going to look at it.
2 Likes
Yeah, its both Problem: FluidNC Crashing Randomly. · Issue #1308 · bdring/FluidNC · GitHub and Problem: Crash and Reboot after 20-30 minutes · Issue #1322 · bdring/FluidNC · GitHub
1 Like
That looks like a different issue. That user was running 3.7.12, where this problem didn’t exist, and was attempting to run his GCode from FLASH instead of SD.
Bart may have been mistaken with the version, I don’t think the bigger memory changes came in til 3.7.18+.
1 Like
that is what I am seeing in github also.
And that is what is reported above as to where problems begin.
but it does not break out the fixes.
2 Likes