Open bipeens opened 8 years ago
Hi @bipeens,
We are working toward adding HA, but currently it is not supported for the control service.
The way you can do this today is by maintaining /var/lib/flocker/current_configuration.json
but also keeping the CA certificates / control certificates safe. When you bring up your new control-service it needs to have the same IP/DNS name as the old one, then things should pick back up.
Thanks! Currently, we are looking forward to this feature. In our side, as nodes can't have same IP/DNS name it becomes difficult to maintain the state.
+1 for key/value store integration for maintaining state
+1 vote for HA
+1 vote for HA
+1 vote for HA
+1 vote for HA
+1 vote for HA
+1 vote for HA
+1 vote for HA
I have started using flocker in my cluster along with mesos and marathon. 1) Can I have multiple nodes running as flocker-control service? 2) If Yes : Is there any leader mechanism in flocker-control nodes? If any node goes down will that affect the cluster? If No: Considering if the flocker-control service node goes down. Is there any mechanism by which I can control such behavior similar to zookeeper in HA.
Also, I tried maintaining the /var/lib/flocker/current_configuration.json for control-service node. But when I try to use for my new created control-service node after the failure of first, I get certification issue.