Observant readers probably noticed that this recipe can essentially be summarized as a combination of the Performing a managed failover and Returning a node to the cluster recipes. In many ways, this is true; it's also the purpose here to emphasize the importance of pausing the repmgr service.
We can't state how important this is: always pause the repmgr daemons before executing potentially disruptive commands. The last thing we want is for the repmgr service to start trying to promote a node while we're tinkering with anything within the cluster. This could produce unexpected results, including service outages!
So after pausing the daemon, it's a good idea to verify that the daemons are paused. Sometimes network disruptions ...
