Duet 2.05 memory leak?

bot

I'm just watching this thread with interest, but have no idea or opinion about why it may be occurring. I was just pointing out the reasoning that octoprint didn't rule out a faulty sd card.

kazolar

@bot I am using a brand new SanDisk 32 gb card. I have used 4 different cards. Are they all faulty?

And the fault magically clears after 40 minutes. And works fine for the next 15 hours...hell I had a night when I caught it in the morning right after it finished a print and was able to run another with no issues. Start cold and the SD card has a fault in it ... I'm software engineer and this makes no sense to me.

dc42

@kazolar said in Duet 2.05 memory leak?:

@dc42 what makes you think it's an SD card error still? I had the same underrun error reproduced with Ocotprint.

This thread is long, so it's hard to find your post in which you mentioned Octoprint (the browser search facility doesn't find it). I recall that you said you had experienced underruns with Octoprint, but AFAIR you didn't say you had experienced the same pattern of underruns i.e. the first print working and subsequent prints not. I replied that underruns when using Octoprint/USB were common on all electronics.

@kazolar said in Duet 2.05 memory leak?:

And the fault magically clears after 40 minutes.

40 minutes of what? The machine standing idle, or something else?

bot

One thought I randomly had last night: Power supplies often have programmed behaviour, based on times or temperatures. Perhaps your PSU is affecting the electronics when a fan kicks in, or something.

kazolar

@dc42 ok, so what's next then. It's warmed up and printing fine at present. I'll be able to get 2 more prints in today, but in the morning tomorrow it will require a failed print to clear the cobwebs again, get it warmed up to get going again...that feels like faulty hardware. Are we still searching for some fix. You tell me it's off warranty. I'll order a new on filastruder today.

kazolar

@bot fan on the main psu is hard wired with no temp sensor in the PSU, and runs all the time. The fan in the PSU runs off its own 12v regulator. The fans for cooling the duet and other electronics are 12v fans running of a separate fanless power supply. duet now, after fixing the z motor connectors is running at nice 26c. So clearly they were underspec'ed and running hot - enough to raise the board temp 3c.

kazolar

@dc42 if I start cold. Even if the machine has been powered for hours after finishing a print. Or off. Same things. The first print using the same procedure - heatup everything, wait for it come to temp, then hit reset start the print. This print fails with high loop times and eventual underruns, that takes 30-40 min. If I do the same procedure immediately following this print, subsequent prints work fine.

dc42

I have just re-read your original post, and spotted something I had forgotten:

@kazolar said in Duet 2.05 memory leak?:

I have done repeated duplication prints without issue -- but this leak happens during triplication -- so obviously more steppers are involved.

I can think of the following ways in which a fault might happen only when doing a triplicate print:

A power issue, if the PSU can't handle the additional heater and stepper motor load. What PSU are you using, and are the VIN terminal block screws still tight? However, your M122 report does not show any power outages, assuming you ran the M122 report just once at the end of the print because M122 resets the min and max voltages.
A temperature issue, because more stepper drivers and heaters are running. However, your M122 report shows a low MCU temperature, assuming you ran the M122 report just once at the end of the print because M122 resets the min and max temperatures.
A firmware issue that is causing DriveMovement objects to be lost from the system. However, your M122 report shows that there were never less than 56 of these free.
Noise on the bus between the Duet and DueX being worse when you drive another stepper motor on the DueX. This could cause I2C data corruption, or possibly spurious interrupts form the DueX to the Duet (which would lengthen the loop time).

None of these explain the increased loop time when this issue occurs. Whereas if SD card reads take longer, that would explain the increased loop time. Also, AFAIR you reported that re-uploading the print file cleared the problem - correct?

PS - it might be better to start a new thread about this issue (linking back to this one), starting by stating what you observe when the issue occurs (stuttering?), what precursors it has (increased underruns? increased loop time?), what conditions are needed to make it occur (triplicate not duplicate?), what has been proven to clear ot pre-empt the problem (M999? Power down/up? Upload the file again?) and what has been proven not to clear it.

bot

What specific PSU do you have?

Is it possible that it has some other behaviour? I know ATX supplies can switch things if no load is detected for a period of time. I am no expert by any means, but I just know that PSUs can behave how they decide is appropriate, and that is not always in line with the uses of them (if they are used in hobbyist projects and not for their exact intended purpose).

kazolar

@dc42 -- I have started a new thread:
https://forum.duet3d.com/topic/15862/duet-2-underruns-high-loop-times-stuttering

dc42

Thanks! I'll lock this one.