Skip to main content

Cluster Add-ons

Atmosly provides 23 add-ons that can be installed on your cluster to extend its functionality. Add-ons are managed through Helm charts and are versioned per Kubernetes version.

Atmosly recommends the following add-ons for a fully functional cluster:

PGL Stack (Prometheus + Grafana + Loki)

A complete observability stack for monitoring, visualization, and log aggregation.

FieldDescription
AppnameName of the stack. Default: pgl-stack
HostnameAccess hostname for the stack. Default: pgl-stack
LokiToggle to enable Loki log aggregation. Default: Enabled
CloudWatch CollectorToggle to enable CloudWatch metrics integration. Default: Disabled
Dashboard Refresh IntervalGrafana dashboard refresh frequency in seconds. Default: 30
Object ExpirationNumber of days after which S3 log objects expire. Default: 90

K8s Dashboard

Web-based UI for managing and monitoring your Kubernetes cluster.

FieldDescription
HostnameAccess hostname for the dashboard. Default: k8s-dashboard

Complete Add-on Catalog

Networking & Ingress

Add-onDescriptionCloudRemovable
NGINX Ingress ControllerRoutes external HTTP/S traffic to Kubernetes servicesAllNo (once added)
Ingress ALB ControllerManages AWS Application Load Balancers for traffic routingAWSNo (once added)
VPC CNIAWS VPC Container Network Interface for native pod networkingAWSNo (once added)
Istio Service MeshService mesh for traffic management, security, and observabilityAllYes

Certificates & Security

Add-onDescriptionCloudRemovable
Cert ManagerAutomates TLS certificate management and issuanceAllNo (once added)
Cert Manager LE EmailLet's Encrypt email configuration for Cert ManagerAllNo (once added)
Letsencrypt HTTP IssuersHTTP-01 challenge issuers for Let's EncryptAllNo (once added)
External SecretsSyncs secrets from external managers (AWS Secrets Manager, GCP Secret Manager)AllYes

Observability & Monitoring

Add-onDescriptionCloudRemovable
PGL StackPrometheus + Grafana + Loki for full-stack observabilityAllYes
K8s DashboardWeb interface for cluster management and monitoringAllYes
Metrics ServerExposes resource metrics (CPU/memory) for HPA and kubectl topAllNo (once added)
Service Monitor CRDPrometheus ServiceMonitor custom resource definitionsAllNo (once added)
HPA for CoreDNSHorizontal Pod Autoscaler for CoreDNS scalingAllNo (once added)

Auto-scaling

Add-onDescriptionCloudRemovable
KarpenterJust-in-time node provisioning based on workload demandsAWSNo (once added)
Cluster AutoscalerAdjusts cluster node count based on pending pod resource requestsAllNo (once added)
Cluster Proportional AutoscalerScales workloads proportionally to cluster sizeAllNo (once added)
KedaEvent-driven autoscaling (scale on queue length, HTTP traffic, cron, etc.)AllYes

Storage

Add-onDescriptionCloudRemovable
EFS Storage ClassAWS Elastic File System integration for shared persistent storageAWSNo (once added)
EBS CSI DriverAWS Elastic Block Store driver for block storage volumesAWSNo (once added)
Single AZ GP3 SCGP3 storage class limited to a single availability zoneAWSNo (once added)

CI/CD & Workflow

Add-onDescriptionCloudRemovable
ArgoFlowArgo Workflows for CI pipeline execution on the clusterAllYes

Node Management

Add-onDescriptionCloudRemovable
AWS Node Termination HandlerGracefully handles EC2 spot interruptions and scheduled maintenanceAWSNo (once added)
ReloaderAutomatically restarts pods when their ConfigMaps or Secrets are updatedAllYes

Add-on Status

StatusDescription
DeployingAdd-on installation is in progress
ActiveAdd-on is installed and running
UpdatingAdd-on is being updated to a new configuration or version
DestroyingAdd-on is being removed from the cluster
FailedThe operation (installation, update, or removal) has failed

Notes

  • Add-on availability is filtered by your cluster's cloud provider and Kubernetes version.
  • Some add-ons have dependencies on other add-ons (e.g., Service Monitor CRD depends on Prometheus). Dependencies are automatically resolved during installation.
  • Add-ons marked as "No (once added)" in the Removable column cannot be removed after installation. Plan your add-on selection carefully.
  • Each add-on version is tied to a specific Kubernetes version. When upgrading your cluster, add-on versions may also be updated automatically.
  • Some add-ons are pre-selected during cluster creation but can be deselected.