Octoprint Will Printer Continue to Print if Connection Lost

I've hit this 3 times on a print in the last 36 hours. Running 1.3.6 on the OctoPi distribution With a minimal X Environment running, but nothing running when the issue occurs. (I didn't even have a web browser installed until I started troubleshooting this issue)

The first time I wasn't able to salvage the print, I think because it had cooled completely and ended up getting layers peeling up. (oddly enough I tried exactly what was mentioned earlier by editing the gcode. (So, I'm glad the logs prints that line out to them. Yes, I do somewhat agree it could be unsafe to automatically do this.) The second time I was able to continue successfully, but was not near the print when the error occurred... until it happened again. I noticed it right away and had top running. and Noticed octopi running near 100% CPU (Image attached) I don't recall doing anything on the PI

I am running a Raspberry Pi 3, dedicated to printing. Nothing else happening on the Pi. I have the Printer (Stock ANet A8) plugged into the Pi's USB ports, along with a USB Hub (Separately powered) , and an additional USB device (K40 Laser Cutter) that was powered off completely via a surge protector (again separate surge protector). In the Hub, I do have a web cam plugged into a USB hub (powered separately) but was not viewing the camera either by a stream of a plug in (control), along with a flash drive and a Wireless keyboard and mouse. Again the USB hub is an active hub and powered separately.

For Power, I have a 5 volt 15 Amp Arcade power supply dedicated to the Pi. I have two sets of 24 AWG (I double checked) cables running from the Supply to power the Pi via the GPIO pins each with separate power and ground connections to the GPIO pins. (They are using headers, I doubt that could be the problem since there are multiples but have considered soldering directly to the board or etching a custom PiHat with terminals. to avoid every possibility.) Someone could say it's providing dirty power. I've tested that as well with my Rigol DS1054Z and it appears to be providing clean power every time I have checked. No I didn't have a scope on it this very time, but have in the past when I've noticed the lightening bolt on the screen to trouble shoot. The lightening bolt also appears to happen when the CPU is under load I noticed.

While we are On power... Jamesh (who appears to be a Raspberry Pi Foundation Engineer) says here talking to the USB device increases power consumption (https://www.raspberrypi.org/forums/viewtopic.php?t=159679 " (e.g. talking to a USB device)")

At any rate the third time I was close by and hear the printer stop after 2-3 hours of running. I had suspected that maybe the printer was for some reason spontaneously rebooting or disabling steppers, so I sat there on my PC in my lab waiting. Oddly enough, the printer hadn't rebooted (a reboot would have stopped the print immediately and as quickly as I turned around I would have seen it happening on the LCD since the screen goes all light and a text "animation" shows as it starts up. None of that happened, it just said "ready". Then I immediately looked at the monitor and the lightening bolt was there and octoprint was consuming 100% resources and python was consuming close to that based on the 'top' instance I had running for the last two hours. (Pic of one window attached) I don't recall doing what I was doing in the second image that did the same thing with respect to CPU resources since the print was working, so I suspect it's a different thing.

Unfortunately, this is ALL I have for information on that instance. The second instance I have more... but it may be a separate issue and I only bring it up here since I saw similar things happening with processes in top the first go round, but I was not uploading gcode the first go round the only thing happening was printing.

I "fixed up" the gcode quickly (5 minutes maybe) and uploaded the file to octoprint through the web interface. The second thing I noticed was the lightening bolt was doing it again. This time I ran 'ps aux| grep python' and noticed that there was a gcode analysis python script running as well. This was eating a large portion of the CPU in addition to the octoprint consuming CPU. I only mention it because if someone was doing that... well, "that's going to leave a mark"so to say. That Python script should be "nice"ly spawned if it isn't, it's a beast it seams.

Whew.... that was a long one. Hopefully, I have provided some additional information or helped in someway to show I'm not some random guy who said "hay muh printes diedz". I've done my homework on this, I'm not saying it's octoprint necessarily-- but I am saying I can't rule anything out at this point. Also, I made a backup of the entire OctoPrint and oPrint folders in case anything there is helpful. Let me know if there is something else I can upload.

At this point, I'm printing that print from SD with OctoPi Connected to see if it does the same thing doing nothing. (I did see it go from 1.1% to 7.0% in top for some reason a bit ago.)

brownfusly1938.blogspot.com

Source: https://github.com/OctoPrint/OctoPrint/issues/1603