Inconsistent Sampling Regularity of Data

I’m getting a behavior that I can not explain and find the root cause of. Attached are two flights from the same hardware and the sampling rate of the uorb messaging is totally different. We think we figured out that the gyro measurement rate drives all the other uorb updates but what can causes the gyro to not update? Bad calibration?

Flight Review - low sampling rate
Flight Review - high sampling rate