Very similar setup to me with VNC, PegasusAstro etc. I save onto a USB3 stick that’s connected directly to the Pi4.

I don’t think I’ve seen any comments from Jasem and others from the most recent crash dumps though.

Ran an imaging session the other night with latest nightly on SM via VNC. Got through lights and darks, but when it came to flats, I forgot to set a decent delay time between images, and Ekos crashed after about the third image. After a restart and setting about a six second delay, it ran the flats sequence.

Seems if Ekos tries to write the very large files created by the 2600 too quickly (short exposures like bias and flats), it crashes.

I'm also seeing crashes which seem related to file size. I'm using the QY268C, which uses the same sensor.
Latest Stable build, using auto-debayer and auto-compute HFR (so I can see these values for focus monitoring).

I suspect the persisting speed of the captured data is slower than the incoming capture data rate in some cases.  The USB memory buffer is set to 256M, but your cameras are producing very large file sizes. If this buffer fills before the image is persisted, there is only room for 1 or 2 images in memory.

To verify this, try running this just before a capture session `sudo echo 512 > /sys/module/usbcore/parameters/usbfs_memory_mb` .  If that gives you more frames captured before crashing, then you may have an overrun.  This doesn't fix the problem, but may show where the problem is, and you will want to find a faster way to persist the image data.

Jim
 

Thanks - I'll give it a try tonight. I'm also wondering if it might be related to this FITSView bug mentioned by KNRO - indilib.org/forum/ccds-dslrs/9059-fitsvi...3c-kstars-3-5-2.html

Whether it is related to the FITS viewer or not should be easy to test. Simply disable the FITS viewer (in the KStars configuration) and see if that solves the crashes.

I just got my ASI2600MC.
Had 4 successful sessions with INDI running on RPI4 (latest SM stable) and KStars/Ekos on Windows client.
No problems.

Tonight, however, I ran into this very problem.
Running everything on RPI4.
After 2-3 frames it crashes consistently with a "unable to allocate memory for temporary bayer buffer: std::bad_alloc" message

OK. Disabled Auto Debayer and Adaptive Sampling - all seems to be good now. Ufff...
Going to a dark place on Friday with my mobile setup - first time with ASI2600.

Has anyone tested whether 3.5.3 brings back the ability to show debayered images?

Or are we stuck with this on these newer camera’s with larger sensors even though devices like Pi’s have enough extra memory in theory to allow this feature to be used without crashing.

It should work fine, but remember we're still limited by the 32bit OS. I think that we don't see 64bit OS before the next stable release of Debian.

I've been having horrible results with EKOS and my ASI2600MC for over a year. I keep hoping software updates will fix it, but all of my imaging sessions are ruined by the PI crashing. I have un-selected debayering in EKOS FITS setup, and it still crashes. Sometimes after taking one frame, sometimes after taking a few.

Very frustrating... I had a ticket opened about this a while back, and some beta fixes were loaded onto my Stellarmate, but it didn't really help.

Scott, can you please submit a ticket to Stellarmate? we'll take it from there.