I tend to agree.
I am normally using the capture module only, and before running a sequence manually start an auto focus run. So the sequence starts without having to move the filter, and on quick look they all have the 'FILTER' entry.
I did spot some missing ones in flatfield images, but don't really have a statistics on that, as I don't store the raw flats for longer periods....
So some race condition in EKOS that depends on history of actions seems a plausible explanation...

So what I found is this: If the default filter wheel is not set, the camera driver is not set to the new filter wheel. In capture module, as soon as you select the filter wheel from the dropdown, it is set a default filter wheel and then this is reflected in the camera --> options --> filter snoop. Is this the issue?

I added a fix that I am testing now so that filter wheel info is also checked whenever a capture happens to synchronize it.

That seemed to be the problem (at least for me) After I clicked on the filterwheel drop down box in the capture menu, the snoop device became visible in the camera xml file. After that filters seem to be set in Fits file all the time.

Thks.

Paul DB

Fix incoming: invent.kde.org/education/kstars/-/merge_requests/397

Hi Jasem,

Thanks for the fix. I will try to do an engineering test later today. I will have to do my own build since I'm on ArchLinux/Manjaro at home.

I'm also seeing this issue with a ZWO EFW and ZWO ASI183MM. Hopefully your fix will correct it.
Is there a workaround maybe changing the filter before kicking off a schedule?

Terry,

Best way around this is to make sure the EFW is in a different spot to what is needed before starting the run.

Jasem,

I tried a local build from that PR branch of yours, still not working. I can recreate the problem.

If you can tell me what I can do to capture verbose logging of what is happening between KStars and INDI that might show what is happening. I have debug logs on but to see the XML requests going through might help.

Cheers
Will

Sorry Jasem, maybe a false alarm. I just double checked the directory where I did the build and the git repo was pointing to master not your branch. I will do a new build and try again. Hold tight

Thanks Will.

Hey Jasem,

Definitely not working on a RPi with the build. I did a complete rebuild using your branch. I tried from both the Scheduler and Capture modules. If I've not used the FW it fails to add the filter on the first images in the capture sequence. After the FW has moved it adds the filter.

I will have to double check on my x86_64 server to see if this exists there or if it's a RPi4 thing.

If you can tell me how I can snoop on the XML requests to the INDI server I might be able to capture those and attach it to this thread.

Cheers
Will

Hey Jasem,

So I can't reproduce this on my x64 Linux server, not even on the public 3.5.4. However, it still doesn't work on the with my RPi 4 running a 64-bit Linux, it sounds like it could be potentially a race condition.

My x64 server is has a AMD Ryzen 5 3600X 6-Core Processor, so it has 12 threads, and it has 32 GB of memory.  My RPi has on 4 GB memory.