Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.jitera.ai/llms.txt

Use this file to discover all available pages before exploring further.

Spot instances (AWS) and Spot VMs (Azure) provide significant cost savings (~60-70% reduction) compared to on-demand pricing. All Jitera application services are stateless and compatible with Spot instances when databases are externalized to managed services.

​
Azure AKS — Spot Tolerations

AKS automatically applies the taint kubernetes.azure.com/scalesetpriority=spot:NoSchedule to Spot node pools. This taint cannot be removed — AKS re-adds it on every node scale-up. The Jitera Helm chart does not include matching tolerations by default, so all pods stay Pending without explicit configuration.
Without the tolerations listed below, all pods will be stuck in Pending with:
0/N nodes are available: N node(s) had untolerated taint
{kubernetes.azure.com/scalesetpriority: spot}
This is Azure-specific — AWS EKS Spot node groups do not apply a NoSchedule taint by default.

​
Required Helm Values

Add the Spot toleration to every component in your values file. The Helm chart does not support a global toleration — each component must be configured individually. 
# --- Spot toleration for all Jitera components ---
# Must be repeated for every component listed below.

automation:
  tolerations: &spot-toleration
    - key: kubernetes.azure.com/scalesetpriority
      operator: Equal
      value: spot
      effect: NoSchedule
  rails:
    tolerations: *spot-toleration
  sidekiq:
    tolerations: *spot-toleration

ultron:
  tolerations: *spot-toleration

ultron_public:
  tolerations: *spot-toleration

boost:
  tolerations: *spot-toleration

litellm:
  tolerations: *spot-toleration

frontend:
  tolerations: *spot-toleration

frontend_core:
  tolerations: *spot-toleration

swef:
  tolerations: *spot-toleration

hasura:
  tolerations: *spot-toleration

html_conversion:
  tolerations: *spot-toleration

playwright:
  tolerations: *spot-toleration

document_converter:
  tolerations: *spot-toleration

kong:
  tolerations: *spot-toleration

# --- In-cluster databases (if not externalized) ---
rabbitmq:
  tolerations: *spot-toleration

redis:
  tolerations: *spot-toleration

mongodb:
  tolerations: *spot-toleration

postgresql:
  tolerations: *spot-toleration

pgvector:
  tolerations: *spot-toleration

# --- Monitoring stack ---
monitoring:
  prometheus:
    tolerations: *spot-toleration
  grafana:
    tolerations: *spot-toleration
  loki:
    tolerations: *spot-toleration
  tempo:
    tolerations: *spot-toleration
  promtail:
    tolerations: *spot-toleration
  otelCollector:
    tolerations: *spot-toleration
The YAML anchor &spot-toleration and alias *spot-toleration avoids repeating the full toleration block for each component. This is standard YAML and works with helm install -f values.yaml.

​
AWS EKS — Spot Configuration

AWS EKS Spot node groups do not apply a NoSchedule taint by default, so no toleration configuration is needed. Pods are scheduled on Spot nodes automatically. For Spot configuration in EKS, see the Amazon EKS managed node groups documentation.

AWS EKS Installation

EKS cluster creation and Helm deployment

Azure AKS Installation

AKS cluster creation and Helm deployment

Sizing Guide

Instance types and resource planning