One customer had a very specific scenario:
They had an Always On Availability Group with Windows Server Failover Cluster Each node had its own storage, so the witness was not disk-based but network-based One node was in the cloud and the other was a physical server The network was unstable, confirmed by a monitoring agent which detected the disconnections, even when the provider assured there were no issues Some network blips caused the cluster to enter the resolving state, with its associated downtime, until the connectivity was restored between both nodes One time, those issues caused the OS to