Hmmm… your description made me investigate deeper. The test I had done with this release (1437089929) was to set up logging temperature with 125Hz sample rate, 32 sample depth, 1s sample interval, for 2 days. It all worked nicely and produced one long logfile.
With the new firmware I set to 8000Hz rate, 1 depth, and no interval and I confirm I see something similar to you. Definitely splitting up the logs. I’ll check this out but for now is it useful to just back off the settings a little? The logs don’t seem to split when I run at 125Hz and 16 samples and the effective sample rate ends up higher than the 8kHz case because the error rate is lower. I hope that helps until I get to the bottom of this edge case.