Quad almost flies away Auto mode, logs attached

Thanks for allowing me to join.

I had an almost fly away on a mission the other day and really skeptical about putting her back up in the air until I have some insight on the issue.

Problem and what was expected:
It was a simple mission hit a couple close waypoints climb up over trees and proceed out to the next waypoint and return. The quad hit waypoint 1 and 2 but started to climb and spiral outward going to the 3rd. I managed to bring it to a halt with stability mode, although I did notice that it had lost the super-simple controls of the stability mode that it had when I launched. It should have continued to climb up to a couple hundred feet and continued to the next waypoint. Stopping the mission gone awry by going to stabilize mode it should have kept the super-simple controls.

DJI 450 frame
Turnigy 980kv motors
Afro 30A Simon K ESC
Pixhawk Controller (3dr)
Ublox GPS and Compass on a mast 6" above props 3dr
3dr 915mhz telemetry radios

Possible Issues?:
These three values in the parameters were marked by Dronshare auto analysts:
INS_ACC3SCAL_X 0.0 Accelerometer3 scaling of X axis. Calculated during acceleration calibration routine
INS_ACC3SCAL_Y 0.0 Accelerometer3 scaling of Y axis Calculated during acceleration calibration routine
INS_ACC3SCAL_Z 0.0 Accelerometer3 scaling of Z axis Calculated during acceleration calibration routine

The GPS Glitch Protection Radius is set to 200.

Looking at # of sat seemed acceptable, at least 10. HDOP errors did seem to increment. I’m kinda new to all this but it seemed the errors climbed when more sat’s were acquired.

Could anyone just point me in a direction, offer some advice or troubleshooting links using the logs.

Thanks in advance for any ideas.

Is the mag interference around 12:02 and 30 sec enough to throw the whole mission off in spiral? And why would going into auto which happened right about that time cause a spike?

I was able to replay in telemetry and HDOP for the GPS was under 2 and Sats were always above 9.

Did you ever get an answer to this?

I am also seeing droneshare complain about INS_ACC3SCAL_X, Y and Z