search

rkaczorek / astroberry-server

Astroberry Server is a ready to use system for Raspberry Pi for controlling all your astronomy equipment
GNU General Public License v3.0
274 stars 19 forks source link

swig/python detected a memory leak of type 'INDI::BaseDevice::Properties *', no destructor found. #181

Open GoateePFE opened 2 years ago

GoateePFE commented 2 years ago

Today I upgraded from Astroberry 2.0.3 to 2.0.4. I could capture darks and lights just fine. When I set bias to 0.0001 seconds (which worked before on 2.0.3) I just get a spinning icon and the biases do not run. The camera is QHY268C.

syslog shows this error over and over: swig/python detected a memory leak of type 'INDI::BaseDevice::Properties *', no destructor found.

I see this same error when running darks and lights, but they work. Bias gives a spinning progress icon with no frames captured.

See screenshot for more details:

image
rkaczorek commented 2 years ago

I guess you use pyindi-client on Astroberry. Unfortunately pyindi-client has been abandoned by original author and overtaken by indi project recently. AFAIK it is not operational as for now. Follow the development and issues here https://github.com/indilib/pyindi-client

GoateePFE commented 2 years ago

@rkaczorek I knew you would have the answer quickly. Thank you. I am not smart enough to change how INDI works in Astroberry. I'm a user. Is there another way I should be using Astroberry that would avoid this issue?

rkaczorek commented 2 years ago

What do you use pyindi-client for? Astroberry does not use it by default to anything.

GoateePFE commented 2 years ago

Please help me understand. I cannot explain. All I know is that bias frames with exposure less than 1 second do not work. When I looked at syslog, those are the messages that I found. I have not changed anything in how Astroberry works.

rkaczorek commented 2 years ago

Oh, I got you. I thought you had gone through a diagnosis that pointed you to pyindi-client - the warning you see in your logs. If this is not the case ignore these log messages (at least for bias issue) and pyindi-client. It is irrelevant. Just to explain - the other topic you have referenced above just says that a user was testing short exposures outside of KStars/Ekos, using a python script.

However it doesn't matter because the actual capturing process and exposure time is handled by indi-qhy driver and related libqhy library (for QHY cameras). These are called by KStars/Ekos when you capture your images.

I have not noticed any issue with short exposures but I don't use QHY. If this is the case with your camera model I would suggest taking a look at INDI forum what other users say. However I don't see any reports on drivers source.

What's the shortest exposure you can take?

GoateePFE commented 2 years ago

Looks like the QHY268C has exposure time range: 30µs - 3600sec. Thanks for the links. I found two threads describing the problem. I just did an update on Sunday, so maybe I didn't get the latest version or maybe there was a regression.

MarticusH commented 2 years ago

@rkaczorek There is a recent thread about this over on INDILib where folks have run into the same problems; happy to help debug further but definitely a problem (this was causing outright crashes for me on smaller-RAM systems eventually).

I don't think it mattered between a ZWO cam or QHY cam, but could revisit to try and help out.