Skip to main content

EKS Deployments

A page for viewing the replica status and availability of Kubernetes Deployments.

EKS Deployments

Key Features​

Stats Cards​

  • Total Deployments: Total Deployment count (cyan)
  • Fully Available: Deployment count with all desired replicas available (green)
  • Partially Available: Deployment count with only some replicas available (orange)

Replica Comparison Chart​

Visually compare Desired vs Available replicas:

  • Semi-transparent cyan bar: Desired (desired replica count)
  • Green bar: Available (actual available replica count)
  • Each Deployment shows available/desired numbers

Deployment Table​

ColumnDescription
NameDeployment name
NamespaceNamespace
DesiredDesired replica count
AvailableAvailable replica count
ReadyReady status replica count
CreatedCreation time

Understanding Replica Status​

StatusDescriptionAction
Desired = Available = ReadyFully normal-
Available < DesiredSome Pods unavailableCheck Pod status
Ready < AvailableHealth check failedCheck application logs
Available = 0All Pods unavailableUrgent action required

How to Use​

  1. Click Compute > K8s > Deployments in the sidebar
  2. Check the Partially Available count from the stats cards
  3. Identify problematic Deployments in the Replica Comparison chart
  4. Check detailed replica counts in the table

Deployment Update Strategies​

RollingUpdate (Default)​

  • Gradually creates new version Pods and terminates old versions
  • maxSurge: Number of additional Pods that can be created simultaneously
  • maxUnavailable: Number of Pods that can be unavailable simultaneously

Recreate​

  • Terminates all old version Pods before creating new versions
  • Causes downtime, used to prevent resource conflicts

Tips​

Partially Available Diagnosis

If Available is less than Desired:

  1. Check Pod status (Pending, Failed)
  2. Check for node resource shortage
  3. Check for image pull errors
  4. Check for Readiness Probe failures
Rollout Monitoring

During deployment, Available may temporarily be lower than Desired. If the difference persists after deployment completes, there's a problem.

AI Analysis

You can analyze with the AI Assistant using queries like "Deployment status", "Find Deployments with replica mismatch", "Analyze deployment failure cause", etc.