Our database was unresponsive for 6 hrs. From our application log, we see a bunch of transaction_timeout. The downtime corresponds to the HA Moving Data graph and seems to be caused by the Role Changes.
Configuration: Redundancy mode - double Storage engine - ssd-1 Coordinators - 3 Cluster: FoundationDB processes - 6 Zones - 3 Machines - 3
The two processes on each machine do not have roles specified.
FDB version is 6.2 (v6.2.20).
Unfortunately, we do not have trace logs at the time. Can you provide insight as to what might be the cause of this downtime? Are we looking at the right metrics?