Last night I wanted to finish a mosaic job, the first half I had done some days ago without issues.

However, this night saw two out-of-order refocus events due to temperature change. After both events image acquisition just did not continue. The first hiatus was ended by the meridian flip. That finished successfully, and after that image acquisition continued. One hour later, another T-induced AF run started, and again acquisition did not continue, sitting idle until dawn made the scheduler shut down the session.

This was with kstars 3.6.1-beta, git commit 419c70156

The first part of the mosaic had been done with 3.6.0-2978_g8b3367d0f. I had started with the same 3.6.1-beta, but had a problem with guiding: The mount suddenly had drifted slowly away in both axes, ignoring guide pulses. Something I've never seen before (I have an CEM60EC with encoders and can easily do minutes unguided if needed). Therefore I had decided to switch back to an earlier version. Last night I had checked guiding during dusk, 48min with total RMS of 0.27", therefore I had decided to stay with the beta this time. Seems that was a bad choice

(I did have some guiding issue again later into the night, sudden huge guide errors/jumps even though there was no wind and excellent seeing, usually the only reasons for bad guiding here)

I read in the git log that there was a major refactoring of the INDI communication - can the be somewhat related?

Attached is the logfile from last night, though it's w/o debug on, so not telling anything (to me)
First refocus was 1:35, MF at 2:44, second refocus at 3:49. Guiding had just continued in both cases, BTW.