Hi, I’m using XtraDB 5.6 with consul for service discovery via DNS and the cluster is failing to bootstrap when there are no nodes yet available.
In my.cnf I have this set up on all nodes, and it works great when the cluster is bootstrapped on one node: wsrep_cluster_address=gcomm://db.service.consul
With the above, I can automatically add any number of new nodes to my cluster without intervention.
But when all three of my nodes are down I get this in my error.log and the cluster fails to bootstrap: “Failed to resolve tcp://db.service.consul:4567” and my cluster won’t bootstrap.
I guess this isn’t totally unexpected behavior, but if I were to leave my cluster_address as empty gcomm:// my cluster bootstraps as expected, and since DNS is returning 0 node addresses I would think the node would bootstrap itself.
Can anyone think of a workaround for this, or is this a bug?
Thanks for your time and input!