Trying to find reason for Toilet bowling

I do understand main reasons for this effect, but in certain cases i cannot able to understand the reason of this behaviour even when there isn’t much magnetic interference, i am using a generic M8N GPS module with compass, it is working well in most cases but sometimes i am getting this toilet bowl effect on my small 5 inch quadcopter build.

For this particular log, EK3_MAG_CAL is set to 3, using HMC5883 compass on GPS module, and also not doing compassmot calibration atm, most of the time same quadcopter seems to be working well without issues, but in this log i am really confused with rare toilet bowl behaviour in guided mode as during most of the flight time it was really stable and quadcopter was just hovering at one point, but during the end, mag x,y,z showing amplifying oscillations, everything was seems to be going well for most period of time but compass had other plans.

In most cases of toilet bowling which i have observed, generally there was already variations/magnetic interference which can be seen from MAG graphs compared to this log before toilet bowl effect start to occur. But why in this log, there was a slow rising oscillation behaviour in MAG graphs in all 3 axis even when there was not much magnetic interference and hardly any variation in mag x,y,z observed from the MAG graphs before toilet bowl effect start to occur?

Log is attached below