Meridian flip issue with the scheduler
sterne-jaeger wrote: Wouter, I think I found your problem. During the meridian flip, the scheduler runs into a timeout:
<code>[2019-03-28T22:48:01.657 CET DEBG ][ org.kde.kstars.ekos.scheduler] - capture module timed out. Restarting request...</code>
The Scheduler does not know about the meridian flip, the meridian flip is handled by the mount module in cooperation with the capture module.
That explains why the capture sequence is started from the beginning and at the same time why there is no re-alignment. The latter is the case since the scheduler thinks that the capture module runs into a timeout and tries to restart the capture.
I have to find out, what the root cause for this timeout is.
Yes that makes sense. At the same time a check is done for the job progress and I think the way I plan my subs for each filter is a problem as well. Or rather, Ekos is not designed for that way. I do series of 9 x L, 1 x R, 1 x G and 1 x B and repeat those all the time until I have enough data. So when the job progress is checked first a check is done for 9 x L, which are found because there are more than 9 x L. Then 1 x R (found as well), 1 x G (found as well) and finally 1 x B (found as well). Then is simply takes the next job which is ... 9 x L and it starts there. But actually this is a restart since the scheduler already passed that point. Just now, for instance, it was supposed to take G sub #3 but it restarted at the second 9 x L.
In any case, I have attached the sequence file and the scheduler file of tonight. Thanks once more for looking into this!
ASI1600MM-Pro Cooled and filter wheel with an 8" TS Ritchey-Chrétien on a SkyWatcher EQ6-R
This is a bug in the meridian flip mechanism of EKOS. The good news: It only occurs when the meridian flip is necessary exactly during the last frame of a job is taken and a second job is waiting for execution.
I will try to fix it asap.
I've posted a new differential that fixes this: phabricator.kde.org/D20151
knro wrote: I've had a weird issue just now as well. Not using scheduler. But I started a test sequence and the capture said "Meridian Flip" but the mount wasn't flipping. I check in the mount tab and meridian flip was checked, so I unchecked and aborted the sequence. When I restarted the sequence, it's always stuck in "meridian flip" state. I believe when the mount module "Meridian Flip" is unchecked, this should probably reset the meridian flip status in capture module (if it was idle at least). The only way to get rid of the "meridian flip" status is by restarting KStars.