Yes, that is weird.

In the log, the exposure aborts first... then guiding.

Also, are you using the ASI1600MM for guiding? The log shows it taking the 4.5 seconds frame.... is that right?

Hopefully, Jasem will chime in here.

Ron

Yes, I have a dual setup, a 2600 on a 140mm refractor, and the 1600 on a 92mm one, for wide field. I use it for guiding when the 140 is the main scope Thanks to USB3 download times are not an issue...

I was checking the git log between the two dates, but nothing caught my attention as an obvious thing to look at. Some hint from the gurus might indeed be needed

Pit, can you explain further about:
    "So something happened between May 3 and May 18 that kills my guiding after a dither"
Are you saying something in the Indi nightly release between those dates, or the KStars nightly release between those dates, or either one?
That is, are you running from source with both Indi & KStars, or running from compiled nightlies grabbed on those days?
When you say, you checked the logs, did you also check the Indi logs? 
Is it possible to reproduce with the simulator?


 

Hej, I thought you enjoy your well-deserved vacation

So to be clear, yes, it is an INDI issue. At least not a KStars one, as I'm currently running the version I compiled June 1 (3.5.4-2083_g95fdd1b37). It was only after I downgraded INDI to the May 3 version that the problems stopped.

This is always self-compiled versions from git. So if there is a suspicious commit I can always check out versions before and after, and cross-check. But if I have no idea where to start....

I just realize that I might also have to re-check libasi. That one was updated to SDK 1.18 on May 18. I had not tried only downgrading that one to the May 3 version. Will do in an hour or so, when the current sequence is finished ;^>

I had thought about the simulator, too. But I'm not very fluent with it. If I find the time, I'll try it, too.

During my Losmandy tests I encountered a similar problem with my 294MC Pro/294MC combo. The guide cam stopped working after taking a video sequence. Since I have not been observing at night I cannot comment on the guiding problem but it sounds very similar to me and the setup (2x ASI) is the same. I also ran the latest GIT versions of Indi and KStars. The issue was 100% reproducible:
Start a video sequence using the guide cam. Stop the video sequence. Then try taking a single photograph. It will result in error. Both cams are "lost" and don't show up in lsusb anymore. After reboot everything is back to normal till the next video sequence.

Gee that sounds bad! But no, no video involved here AFAIR(*). And guiding/taking images works flawlessly for hours - if I don't dither
Yes, I'd like to test some other camera, but ZWO is all I have (4 of them....)

(*) I regularly do use video for the main cam, for initial focus.

I did a fresh git pull and now I am unable to reproduce the issue. Don't know wheter or not it has anything to do with the fresh git pull. In any case it had hampered my tests for several days. Should it re-emerge I'll post a video. Weird!

Hi Pit, everyone

FWIW and as Hy asked, it's working OK with the simulator from git a few days ago:
kstars: be4c957b6e971acac17f325fcac796e20e864cfe
indi: 3b4d0fe80c821bd1241e1dc801550dcd6f19d097

Logs attached.
Try the latest git?

Anyway, HTH and clear skies,
Steve


 

Hi Steve,

thanks for the check!
I have meanwhile verified that it is not the version of the ZWO SDK. I upgraded that one to 1.18, and didn't have issues last night.

Just checked the indi git, and there haven't been relevant changes since June 1. So I doubt that would change things.... I'll try to verify the simulator also works here, that should at least narrow down the issue a bit.

So I compiled todays git for both indi and kstars.  As I was afraid, no change.  As soon as dither is finished and guiding is resumed EKOS (or INDI) start aborting the ongoing guide exposures:It seems to me a command is sent to capture a frame, although guiding has never been stopped, and an exposure is already going on!?
I tried twice, same error, then reverted INDI back to the May 3 version (no downgrade of kstars.  So either it is fully within INDI, or some protocol has changed in INDI that EKOS doesn't comply with?

I checked the successful logs.  It seems, also with the old INDI version the first exposure after the guiding is resumed truncates the ongoing exposure, if I look at the timestamps:That is only 2.7 seconds between the two messages 'taking a 4.5s frame'.  But there it's only for this single moment.  All other exposures are taken properly, whereas for the new INDI it looks as if two processes were trying to expose, and they abort the exposure of the other....
(Sorry if I sound weird - just typing what comes to my mind...)

If someone has a smart idea how to narrow this down, I'm all ears!

 

Thanks for the report. I'll test tonight in the observatory and see if I can replicate this.

Video and logfile of the ASI guide cam freeze. After a while both ASI cams are lost and don't show up in lsusb anymore. All other devices are OK. All on latest GIT.

The logfile says:
[2021-06-06T00:59:16.886 CEST INFO ][           org.kde.kstars.indi] - ZWO CCD ASI294MC :  "[INFO] Starting the video stream with target exposure 0.100000 s (Max theoritical FPS 10) "
[2021-06-06T00:59:32.015 CEST INFO ][           org.kde.kstars.indi] - ZWO CCD ASI294MC :  "[INFO] Starting the video stream with target exposure 0.500000 s (Max theoritical FPS 2) "
[2021-06-06T00:59:32.016 CEST INFO ][           org.kde.kstars.indi] - ZWO CCD ASI294MC :  "[ERROR] Failed to set exposure duration (ASI_ERROR_GENERAL_ERROR). "