Lost connection when uploading Gcode file during print

MartinNYHC

Just had the connection reset when uploading a file but not printing.

Phaedrux

What kind of SD card are you using? Do you have an A2 class to try instead?
How large is the file you're uploading?

MartinNYHC

I'm using PXE boot with GigabitEthernet on a very performent NAS.

ls -alh carrot_patch_red_set.gcode
-rw-r--r--@ 1 xxx.yyy  staff    44M Jul  6 20:05 carrot_patch_red_set.gcode

chrishamm

These two lines

Jul 06 16:08:33 Voron2 DuetControlServer[6350]: [info] System time has been changed
Jul 06 16:08:33 Voron2 DuetControlServer[6350]: [warn] Controller has been reset

indicate that the IO throughput is too low and/or the CPU load too high. If you really need to run your Pi from a PXE source, I suggest you copy DSF to memory (e.g. /tmp) to keep the latencies as low as possible.

MartinNYHC

thanks. ok, makes sense.

can you please explain what you exactly mean with "DSF"? the complete folder? what i'm thinking about is the fact the the "gcodes" foder si also a subfolder

chrishamm

@martinnyhc I'd start with the /opt/dsf/bin folder and check if that improves things. If you have macro files which are called frequently from G/M-codes, I'd copy the system folder as well (/opt/dsf/sd/sys).

PS: It might be possible to set up cached folders in your PXE system config (comparable to offline files on Windows) but I can't say if that is possible. Might be worth checking though.

MartinNYHC

allright, that helps. will try things out and let you know.

MartinNYHC

All of a sudden I get the following error message:

Jul 07 16:39:38 Voron2 DuetControlServer[952]: [warn] IPC#14: Client with outdated protocol version connected (got 7, want 11)
Jul 07 16:39:40 Voron2 DuetControlServer[952]: [info] Received file abort request on channel HTTP for all files
Jul 07 16:39:40 Voron2 DuetControlServer[952]: [info] Aborted macro file homeall.g

Any ideas?

chrishamm

@martinnyhc There should be something above that log part showing the reason for the file aborts. Possible reasons may include bad probe readings, emergency stops, stack overflows (by M-codes or M120/M121), or G-code exceptions (e.g. from bad expressions).

That first warning message from your excerpt comes from a plugin that doesn't use the latest API version.

MartinNYHC

@chrishamm said in Lost connection when uploading Gcode file during print:

@martinnyhc There should be something above that log part showing the reason for the file aborts. Possible reasons may include bad probe readings, emergency stops, stack overflows (by M-codes or M120/M121), or G-code exceptions (e.g. from bad expressions).

That first warning message from your excerpt comes from a plugin that doesn't use the latest API version.

Yeah, in the meantime I've found out that the Z endstop was the cause. Replaced it and everthing's good again

Will check the SBC tmpfs thing today and let you know the results.

MartinNYHC

Moving dsf/bin to tmpfs didn't solved the problem. But that's fine. Just don't upload large files during a print. Need a reminder

MartinNYHC

I decided to use a SD card again and only mounting the macro and sys folder via network. That should solve the issue and I don't need to take care of backing up these files.