Familiarize yourself with requirements and results before you drop a multiplex server.
If the target server is running, it is strongly recommended that you shut it down before dropping it. In case you do not, it will eventually shut itself down, but it is better that you plan for and shut it down prior to dropping it. You cannot drop the coordinator node and the designated failover node unless it is the last secondary node. When the last secondary server is removed, the multiplex is converted back to simplex and the coordinator shuts down. A write server that is holding free list cannot be dropped.
A normal restart of the write server will give up its free list and then you can shut it down and then drop it. If, for some reason, the writer cannot start, and you can guarantee that its process is dead, you can restart the coordinator with the -iqmpx_reclaimwriterfreelist switch. This forces the coordinator to reclaim the writer's free list and it can then be dropped. This startup flag should be used only when you can guarantee that the writer process is dead and cannot be started. If the writer process is still writing to the database when the coordinator reclaims its free list, database corruption may result.