Manual scaling
Search workloads, inspect current and desired replicas, and submit a focused replica change from one dialog.
Fast Kubernetes scaling · interactive mock
Scale workloads with operational context intact.
Scaler explores a focused control plane for manual replica changes, second-granularity reconciliation, schedules, suggestions, operation timing, and timeout budgets across Kubernetes workloads.
React 19KubernetesEKS / IRSA
boltScaling overview
eks-prod-01Active replicas
38
6 pending
Median scale
1.9s
↓ 0.4s
checkout-api
CPU · 61%
events-worker
SQS · 8.2k
pdf-renderer
Schedule
Control-plane model
A faster scaling workflow, end to end.
One-second reconciliation
The interface models desired-state reconciliation every second and makes scaling state visible per workload.
Schedules and signals
Bring CPU, request rate, SQS depth, schedules, and recommendations into the same operational view.
Timing visibility
Track last-operation latency, median scale time, success rate, and consumed timeout budget.
Local proxy boundary
The documented access path uses local kube-proxy and explicitly warns against publishing it through ingress.
Credential choices
Supports node credentials, recommended least-privilege OIDC/IRSA, and static keys only for local development.
Run locally
A realistic interactive prototype.
Docker Compose starts the interface at localhost. Its controls are interactive and use representative workload data, but the current project deliberately does not modify a real cluster.
$ docker compose up --build
open http://localhost:3000
Security posture
Least privilege before automation.
The design recommends IRSA for EKS, an IAM role limited to configured trigger sources, and Kubernetes RBAC constrained to the intended namespaces and workload kinds.
Static AWS credentials belong only in local runtime secrets—never in the image or repository.