Comment 2 for bug 1580817

I think this happens when the rack loses its connection to the region while looping in NodePowerMonitorService.query_nodes(). It's not critical at all, but I'll improve the error message.