Assume that you use an Always On Availability Group in SQL Server 2017 on a Pacemaker cluster. If the cluster tries to fail over the primary replica, the cluster may not demote the current primary replica because of an issue such as a connection timeout. In this case, the cluster ignores the failure and continues to promote another replica to primary. This results in two SQL Server instances that are the primary replica of the availability group .
Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.