Debugging EKF variance error

During an indoor flight my quad experienced an EKF variance error. The drone entered LAND mode but it started drifting slowly and crashed with a wall. I am trying to figure out the sensor(s) that led to the EKF error and whether this could have been avoided in some way.

You can find the log here: Dropbox

EKF variance indoors always makes me nervous since GPS is already degraded and the filter has fewer sources to reject bad estimates. Checking the VIBE and IMU log messages against the variance spike timestamp usually points to the root sensor pretty quickly.