There has been an hour-downtime yesterday evening (07/02/15). For transparency reasons, we want to explain what caused the outage.
Since last week, one of our hypervisor's power supply was in troubles. We discussed with our hosting provider and we agree to change it at 3:20 p.m yesterday.
However, we have been notified at 6.05 p.m that the component will be replaced in the next minutes... ie approx three hours after the scheduled time.
At the same time, a router in the network of our hosting provider went down, and instable. A node in our Galera cluster has been impacted and began to switch between reachable and unreachable very quickly. This behaviour affected the cluster synchronization.
Due to this instability, the website was unavailable. As soon as we found the cause of the problem, we isolated the affected server to make the cluster correctly synchronizes again.
When the router became stable again, we reintegrate the server in the cluster.
All of this has lead to a downtime of approx one hour, between 6 p.m and 7 p.m GMT+2.
Since, everything is normal again. We are very sorry for the inconvenience.