Adjusting the scale doesn't take the shortest path #1188
Labels
kind/feature
Categorizes issue or PR as related to a new feature.
priority/important-soon
Must be staffed and worked on either currently, or very soon, ideally in time for the next release.
triage/accepted
Indicates an issue or PR is ready to be actively worked on.
Is this a bug report or feature request?
What should the feature do:
The operator should stop scaling up if a new desired state has less members.
What is use case behind this feature:
Say we have a scylla cluster with
3
nodes and want to adjust members to5
but accidentally put55
there. The controller will create55
services which is fine but when we correct ourselves and change the members to5
, the controller will first scale up to55
nodes and only then goes back to5
. In the worst case your infra won't be big enough to scale to55
instances and you'll get stuck.Additional Information:
We are creating the services ahead of time which is fine but we need to make sure we stop scaling up as soon as we can and somehow differentiate the services that never had a bootstrapped node (which may be hard).
The text was updated successfully, but these errors were encountered: