[arielr-lt]

What

Set ng_staging_min_size = 0 (was 1) in terraform/environments/eks/terraform.tfvars for the ce-registry-eks cluster.

Why

The ce-registry-eks-ng-staging node group (1× t3.medium) is empty — there is no staging namespace on the cluster and nothing targets its env=staging taint. It was floored at min=1, so the cluster-autoscaler could not reclaim the idle node. Cost cleanup alongside the Argo/ES/Neo4j decommission node-group reclaims.

How it works

• desired_size is under lifecycle.ignore_changes (autoscaler-managed): Terraform leaves it at 1, and the cluster-autoscaler drains desired 1 → 0 once min_size = 0 applies.
• max_size stays at 6.

Safety (verified)

• The sole node runs only DaemonSets (0 app pods), tainted env=staging.
• No staging namespace; 0 pods anywhere target env=staging.
• Only the staging node group is affected — prod (ng-prod-v2), sandbox, and private node groups are untouched (separate resources/variables).

Rollback

Raise ng_staging_min_size back to 1.

Fallback (if autoscaler is slow)

aws eks update-nodegroup-config \
  --region us-east-1 \
  --cluster-name ce-registry-eks \
  --nodegroup-name ce-registry-eks-ng-staging \
  --scaling-config minSize=0,maxSize=6,desiredSize=0 \
  --profile credreg-sso

[rohit-joy]

@arielr-lt Is this needed? If not, please close PR.