This should at some point be configurable and maybe even allow autoscaling.
That's not to ensure load capacity but availability. Take for example our staging & dev clusters where the nodes restart every couple of ours and so. During these kinds of periods, a single instance might become unavailable and we would not want that.
Maybe we could get some actual usage of the controller first, but for future record.
At this stage helm chart allows to configure number of replicas. Later we could expand it (depending on real needs) to allow configuring autoscaling.
As mentioned in this https://github.com/kubeshop/monokle-admission-controller/pull/8#discussion_r1365228665:
At this stage helm chart allows to configure number of replicas. Later we could expand it (depending on real needs) to allow configuring autoscaling.