I had a production redshift cluster that underwent its scheduled maintenance and its queries took 10x longer afterword. I had another production m1.large which

I had a production redshift cluster that underwent its scheduled maintenance and its queries took 10x longer afterword. I had another production m1.large which mysteriously started hitting 100% after maintenance - despite the fact that the other instances with the exact same software stacks were fine. In both cases I had to recreate the instance/cluster because I had no idea what to root issue was