Good morning, I would like to try to better to understand how the load balancing mechanism for clustered nodes.
In my situation I have two nodes, one of which is a leader while the other worker. When do I get to a service using the IP of the leader node. If my leader node becomes unreachable, as I reach the services? Swarm switches the Leader IP node on Worker node?
Swarm with one manager and one worker cannot accept any manager failure. You need at least 3 managers, that one manager is allowed to fail.
So in your case if your leader/manager dies, your swarm is dead. Worker will not be promoted to a manager.
Ok. If I have tree docker host with Swarm. Two container as leader of cluster and one si worker. If one of two leader became down and I reach service from IP of this leader, what happens?
Swarm migrate IP in another leader host available?
Here who options you have i guess . because of we are talking about not application health instead health of worker and manager nodes its easy .
Manually you should put high available reverse proxy in front of docker swarm for example nginx , ha proxy , netscaller , f5 , any
Other way DNS but not static one for example azure traffic manager or amazon route53 which is capable to monitor ip address and if one of down then change dns entries dynamically .
VM