There was 7hrs downtime caused by wrongly configured igmp queries within multicast vxlans during migration to the new hardware.