Dear Colleagues,
I have noticed that when the leader node goes down or leaves, it takes the cluster about 16 seconds to elect a new leader. During this period, local Consul clients cannot use the services of the Consul cluster (the K/V storage etc), with error=“rpc error making call: No cluster leader” and such.
To my mind, 16 seconds of unavailability is too much. For example, Patroni is not happy about the outage. Have I misconfigured something?
If the 16s re-election outage is unavoidable, how do I manually shift the leader role to another consul server before I can stop or take down the current leader?