APP Crash during Calibration

Such a hard crash is pretty rare, usually this would point to a system issue. What system are you using, version, amount of memory etc?

Relevant system specs are:  Intel i7-9700K @3.6Ghz (8 CPUs) - 16GB RAM - Nvidia GeForce RTX 2080 (16GB)

However, I decided to run the process again and this time I noticed that the "Multi-Channel/Filter" and "Multi-Session" options were checked. Because my capture session used a Canon 80D DSLR, I unchecked these options first. This time it completed as normal. Not sure if those selections were causing the problem, but I can report any future occurrences if they happen.

Thanks

Thanks for that, that's interesting for me to try as well. Might just be coincidence though as I've not seen this mentioned before.

Well, it happened again just now. I loaded all my lights, darks, flats and bias frames from a capture session last night. Started the CALIBRATION step, and after several minutes of processing, APP crashed and vanished. No error message of any kind. This time I had the "Multi-Channel/Filter" and "Multi-Session" options unchecked. If there is anything other than my PC specs I can provide that could help you figure this out please let me know. But I'm beginning to think I may have to move to another processing program. I'm 2 for 2 at this point...

...as a follow-up, I tried the processing again and the same crash happened - not sure if it helps but I saw that it happened during the processing of my flat files. They all look normal so I don't think there is a corrupt file...? Not sure what to do at this point 🙁

...additional follow-up; for this project the crash is persistent, but it happens at various points during the CALIBRATION process. If it helps to troubleshoot the problem I can post my capture files to a cloud drive so you can try it yourself.

9/19/20 ...Success! I ran the process again today but this time I left all CALIBRATION options at the defaults (I was using "Average" for Integrate and "Adaptive Rejection" for Outlier Rejection as recommended by the tool tips.) Hopefully this provides some insights for you, and if you need my files to work on this problem please let me know. I would like to be able to use the program as designed. 

Very strange, the only time I regularly had a crash that big, was due to me overclocking my workstation (no other program crashed, but APP is the most taxing on the system). So that was due to a problem in the system and it's stable as a rock since I fixed that. You could try to launch APP from the terminal (or shell) as that's where the system might output some crash data even in this hard case. I would still advice to do a good check on your system as well with some stress software.

I captured another project last night. Loaded the data into APP today and started the CALIBRATION step with the defaults as before. Unfortunately the program crashed again. I run other high performance software without any problem (CAD, 3D Studio, etc.) so I have serious doubts that there is something wrong with my system. But if there is I'm not sure how a stress test would identify it. I've run those kind of tests before without issue, but I'll see what I can do to test my system further. Otherwise, unless there is a refund policy it looks like I may have wasted my money. No hard feelings though... not much you can do remotely I suppose. 

I'm on 1 month trial for now - so no money involved yet - but I experience the same behaviour.
Sudden crashes with no error message. Just vanishing to desktop.
Probably all of them during calibration - sometimes at the very start, sometimes near the end.
Tried like 10 times and had no luck with finishing any run.
Buffed graphics PC -16gb RAM, core i7, 250+gb free space on working HDD.
Often running 3ds MAX, Photoshop and Corel Draw simultaneously with no issue at all.

Edit: I remembered that twice it passed calibration stage and crashed during stars analyse.

@dave23913 Oh, we definitely would like to know what is causing it, this shouldn't be your experience of course. Launching it from a shell or terminal might give a bit more information on the crash, if you could try that? If that doesn't work, I'd like to have some set of data so I can try to reproduce it.

@mikuda We haven't seen these type of hard crashes much, like I mentioned it's mainly a system issue usually or something with the graphics card (some very new ones misbehave I believe regarding the driver). Could both of you state your model of gfx card? And terminal/shell output would give some info as well hopefully.

Win7 64 bit
Nvidia GeForce GTX 970 
Can You give me quick instructions how to run it in shell?

My system is Win 10 Pro 64 and my GPU is Nvidia GeForce RTX 2080 (16GB). I have raw data I can send but will try terminal launch as well. 

Ah yes, I'm afraid there might be still an issue with these cards. Does it work when you switch off OpenGL by clicking the GL button on top of APP?

YEA! (for the moment)... After turning OpenGL off as you suggested I was able to run my project from beginning to end without issue. Seems like the fix but only time will confirm for sure. Thanks for sticking with me to figure this out. Here's my initial project result of IC1318:

40 Light frames 1600 ISO 240s
40 Dark frames
40 Flat frames
40 Bias frames
Additional post processing in Adobe Photoshop

Excellent, yes that's likely the fix (though it will run slower) and indeed an issue with the modern drivers from NVidia. Mabula is aware of that, I'm not 100% sure we can completely fix that on our end, might take the migration we're planning towards a more modern Java version. Sorry for the inconvenience.

No worries - I didn't notice the extended processing time - just happy it now works. We've had three consecutive clear nights here in Ohio. It's 3:27 am now and I'm finishing up a capture of Andromeda. Anxious to see how that turns out. Thanks again. Cheers.

Wow, hardcore in the middle of the night in the middle of the week! 😉 Good luck!

FYI: My M31 project failed today with OpenGL turned off, but I went back and used the default options for the CALIBRATE and INTEGRATE processes (instead of "average" and "adaptive rejection"), and it completed as normal. I hope this provide some additional insights for you. As long as I can get through the process I guess I'm good because I always do additional post processing in Photoshop. 

120s @1600 ISO
87 Lights (had three bad images)
90 Darks, Flats and Bias

Hi! I'm posting late reply since I wanted to try "OpenGL option OFF" attempt first. Can confirm that it passes whole integrarion now!(At least with default settings - considering Dave's last post).

I can't compare it to "openGL on" process (since I couldn't finish any) but whole integration (all 6 steps) took really long time. Few hours at least - and it wasn't especially big integration. I'm fine with that in terms of "waiting for results" but I can't afford to block my computer for so long. So, how can I "save" the process after each step to continue it later?

Also, which options I can fiddle with to try to speed it up?

Thanks!

@vincent-mod

Ok so I Have the following situation:

PC is an AMD 32 cores Threadripper (64 threads) with 256Gb of RAM (don't ask what I do for a living)

Two display adapters : Nvidia GT 710 and one Nvidia RTX 2080 TI. <- all driver to the latest version 

Windows 10 pro up to date 1909 last rev with all the updates applied

First problem has to do with the display of the APP windows.  On one of the two monitor I use the windows are totally wild and any movement of the mouse is producing the bug.  On the second monitor all is fine with the display of the APP windows. 

Since you hinted issues with the Nvidia hardware I am mentioning this since it might have context for you.  I turned off the OpenGL option but it does not fix the display of windows.

Now for the real problem 

I am trying to integrate 4 session with differents exposures with all their own Darks, Flats, DarkFlats, and Bias files.  

In total I have 

152 Lights

153 Flats

54 Darks

151 DarkFlats

103 Bias

No matter what I do I get an application crash at calibration:

------------------------------------------------------------------

Encountered error in module:
CreateCalibrationDetails

Error message:
java.lang.NullPointerException

Cause:
null
bV.a(Unknown Source)
bV.doInBackground(Unknown Source)
javax.swing.SwingWorker$1.call(SwingWorker.java:295)
java.util.concurrent.FutureTask.run(FutureTask.java:266)

----------------------------------------------------------------

I usually just stack simple project and so far I was enthusiast by APP.  This is the first time I take a project this size to APP and I must say I am scared at the application stability.

Thanks for considering this as part of your investigation

Well, first off; never be scared of software. 😉 But yes, that's an unfortunate situation. I will ask Mabula to have a look at this thread as well to see if this is indeed the issue we have with some of the newer cards.

@vincent-mod

Thanks for taking car eof this Vincent.  Not afraid of software... After a Bachelor in physics (B.Sc. U of Montreal), I have been making a living at software for over 30 years.  Soooo I do have to tell you I have the outmost respect for the algorithmic and scientific value of APP !!!  Once you do a couple of tutorials to unfold some of the mysteries, then it is obvious that APP will be part of my life !

More info:

I was doing this calibration on v1.082.

I have Java RTE 1.8.0_251 ... I know how that can influence the context.  I could go to 8 update 261 but I am worried for other software I use.  Unless to tell me this would be fixing things, I am not too hot on updating.  What version of Java is your test environment ?

I also have looked into the security definitions in Java to make sure I was not preventing a module from executing.  Not the case.

On a fresh start, after a reboot to make sure all is clean, I can add lights and calibration frames up to a point where the software starts behaving strangely.  

Screens are erratic and it is obvious something is wrong with lists or pointers in Java memory (geee I have seen this too often in Java).  Afterall, that is the error I get.

As long as I do not get to that point, All is fine and I can calibrate.

So I ran the following experiments:

1) I did the project in chunks to make sure there was not issue with the data.  And Yes that worked.  4 session done separately one by one did work properly.  (that gave me 4 final fits file to later integrate.  But for large project that is an intense workload I would like to avoid.  So that sounds like the mosaic project in Mabula's tutorial.

2) I played with a lot of the parameters to see if this was influential.  No it is not.  In particular, OpenGL or not does not make a difference.

Again thanks

Claude

Java on your system is of no influence, but the cause of this does have something to do (we think) with us using Java 8. So we're going to switch to Java 11 and tackle this issue then as well as it's definitely not what we hope to see.