The explanation is that in the fourth iteration the third job's autofocus failed.
Therefore this job was marked as aborted and not considered any more.
To prevent such cases, you need to change the aborted job management from "None" to "Immediate.

Thanks for the feedback, will update the jobs and give it a new try tonight.


The rest in our log looks good, Parts 1, 2 and 4 are cycled through as expected.

That I do not fully understand.

Job 1: 15 + 15 + 15 + 15 + 15 + 15 + 15 = 105 lights.
--> Seems ok to me, because after the last series of 15 lights the next job is started.

Job 2: 15 + 15 + 15 + 15 + 15 + 15 + 15 + 14 = 119 lights.
--> This job should have stopped after light 105 similar to job 1 and moved on to the next one.
--> But 14 too many shots are taken, after which the astronomical night was over and the rig shuts down.

Job 3: 15 + 15 + 15 = 45 lights.
--> This job was stopped after 3 runs due to clouds that prevented focusing, so that's explainable.

Job 4: 15 + 15 + 15 + 15 + 15 + 15 = 90 lights.
--> If job 2 had been stopped correctly, the 14 lights would have been added here.


Am I making a reasoning error?

I'm not 100% sure, since I cannot find an explanation in the log. Could it be that Part 2 is the last one that could be executed before dawn? All others seem to be postponed to the next night.

I don't understand it well either, but based on previous tests it seems to me that the scheduler no longer starts the next job if it can not be completed completely. And then the current job continues to repeat until dawn.

Maybe I'm wrong, but if so, it has little added value, especially with mosaics. Because you actually want to have approximately the same number of lights of each tile and each (different) night.