Stange MK3S/MMU2S software bug in the middle of a print
I didn't catch the terminal in time to see the actual debug logs, but I ran into a really strange glitch on a print today. On most current firmware, 3.7.1 + 1.0.5.
Was printing single color (in MMU2S mode) part and about halfway through it stopped for no apparent reason. I went to check on the printer and it acted like I had wanted to change filament, it had moved two positions right and stuck the filament out of the MMU2S body by about 10cm, told me to replace the filament and press the button. I clipped the tip and pressed the button and it resumed, only to retry about 20 minutes later.... always on the same spool with no extra user input, and moving to the front to dump the excess filament when I "changed" spools.
I get to a third time, do the same thing, and it never manages to leave the "Loading Filament 1" screen, even though it was in the middle of the print using a single filament with no tool changes. Pressing the middle button on the MMU2S advances the filament slightly, returns it, and returns to a solid green light. Filament select buttons work fine on the MMU2S.... but the thing just would not resume. After a variety of inputs, it eventually moved on top of the half completed object and tried to redo the reload thing - dumping extra plastic into the part, and then continued from the position like normal, with the now unusable part (it was a spring...).
I tried to remove the blob before it set and was largely unsuccessful, causing a crash to be detected with the extruder. After the crash and the head moving out of the way, I moved the Y axis unintentionally and heard the skips happening, but the printer started trying to reprint in a different location with the Y axis slipped... so I hit the reset button and made sure everything was stopped in octoprint.
Now this is definitely a software bug, even if something was causing the retries on loading filament, it should never have been allowed to try to dump filament as a purge for a new one being loaded directly into the part. A lot of my experience with the MMU2S firmware so far has been a bit beta-y, there have been a lot of instances when it's clearly doing something or noticing something and it doesn't communicate it to the user, plenty of times when it's got caught in a loop and seems inactive in doing loading/unloading operations when the sensors aren't reading out properly (which has been a bit, since the IR sensor is so difficult to nail the calibration on), and now a clearly incorrect behavior actively ruining a print. For this specifically, I think there should be a sanity check before running that reload purge gcode, but more generally, more feedback on the screen as to when something is happening or when something isn't acting properly would be greatly appreciated.
Also worth asking: the spool being used was almost run out (I can't reprint the part until I get a replacement as I don't have the 10m left this print requires in the filament I want to use), and there was probably more friction in the PTFE tubes than usual because of the tighter bend. It seemed like it was fine pulling the filament into the MMU2S, but can this cause the printer to think the spool is out and needs replacing even before the end of the filament passes by the FINDA?