Every few weeks we are getting into trouble inside some clustered service hosted inside the datacenter of one of our hosting providers. Trouble means: service cluster members ar