I've read this guide about cluster architectures, and I've built a development cluster with this config:
All machines are on the same subnet.
One machine acts like a master, and runs etcd only (IP address is 192.168.0.95).
#cloud-config
coreos:
etcd:
discovery: #url
addr: $private_ipv4:4001
peer-addr: $private_ipv4:7001
units:
name: etcd.service
command: start
I then have three workers that run fleet with this config:
#cloud-config
coreos:
fleet:
etcd_servers: "http://192.168.0.95:4001"
units:
name: fleet.service
command: start
I can control machines and units via fleetctl, but I haven't been able to use etcdctl properly.
Each worker uses local etcd when working with keys, therefore keys exist on the worker they've been created on only!
Isn't this supposed to happen, given the above config?
Thanks
You'll need to point
etcdctl
at the etcd cluster just like you're doing with fleet, right?etcdctl --peers "http://192.168.0.95:4001" ls /