We have a 6 node elastic cluster on bare metal that was installed by way of helm 7.10.0
We are looking to downsize the cluster from 6 nodes to 3. The 3 that are being decommissioned have the master pods on them along with data pods.
Are these steps sufficient in moving those pods off those nodes and having them rescheduled on the 3 nodes that are schedulable?
Are there any steps needed for the pv's | shards etc?
- cluster.routing.allocation.enable" : "all"
- Taint the nodes noschedule (nodes being decommissioned)
- Rolling restart of master statefulset
- Rolling restart of data statefulset
Thoughts??
Thanks so much