Bad EKF position solution

We ran into a problem with our big hex with hybrid engine today after several successful flights in the past. The log for today’s flight is here.

We armed the hex and turned on the hybrid engine but disarmed due to gust but did not do a power cycle. The hex stayed on for ~30s then we armed again. Switched to Loiter and it starts freaking out and we had to land. Looking at the log, it seems like we were having bad ekf estimation of lat and alt (lon looks fine) and it seems to be accumulating while we were on the ground. Unfortunately, we did not enable LOG_DISARMED so we do not have useful data during the landed phase. We would like to see if there are any log field that might be useful to check to see what could be the issue from this log. We will try to replicate the problem with the LOG_DISARMED enable as well but want to try to recover any useful data. For more context, our Hex uses 2 ardusimple f9p in moving baseline configuration which has worked well. We have the ardupilot configure the GNSS module during startup according to the tutorial on the wiki and the moving base transmite RTCM to the rover via a direct wire connection throught the GNSS UART2 instead of through the Cube.

On another note, on Mission Planner map page, does the vehicle position on the map based on GPS position or the EKF solution? We took off with bad EKF because the position on the map looks right. Is there a way to have the EKF error or flag be more pronounced on the Data page of mission planner? Thank you for your assistance

Is it ArduCopter 4.4.4? You should use at least that version.
What you seam to describe is a bug that got fixed in 4.4.0

Thanks for the response. We will look into upgrading. Would you mind sending me a link to the issue in 4.4.0 that you are referring to?

It’s on the release notes