On Saturday June 20th around 10:40am PDT, two Virtual Machines running database containers experienced extreme unresponsiveness and ping failure. This affected 0.2% of sites on the platform. Upon alert, all sites with Disaster Recovery enabled were failed over to their replica database. These sites were unaffected for the remainder of the outage. After recovering the virtual machine it appeared stable for an hour before networking began failing again. At this point we proceeded to replace the virtual machines and recovered all sites. The total on and off disruption went on for approximately three hours. We are currently working with our service provider to understand the nature of the network failure to these servers.