Waiting too few may result in the image not having finished being pulled, therefore the migrations not being applied since the old pod is still running.