Job freezes and dies if too many runs/histograms are used at once. We should look into why this is the case and how we can improve this.