Comment 3 for bug 1469621

Running resolved --retry in turn on each unit will resolve this problem; alembic is clever enough to figure things out.

Fix proposed to ensure that db upgrade is only run on the lead unit.