AKS Node Pool Topology

This section explains the AKS node pool configuration differences between active (primary) and standby (secondary) clusters.

Node Pools by DR Mode

The following table shows node pools by DR mode:

Node Pool Primary (dr="active") Secondary (dr="standby") Purpose
System Deployed Deployed AKS system components
Digest Deployed Deployed Core LogScale processing (queries, indexing)
Kafka Deployed Deployed Kafka broker nodes for message queue
Ingress Deployed Deployed Load balancer and ingress controller
UI Not created when dr="standby" Not created Web UI serving
Ingest Not created when dr="standby" Not created High-volume data ingestion

Node Pool Creation Logic

The UI and Ingest node pools use the following Terraform count condition:

terraform
# UI node pool (logscale-ui-node-pool.tf)
count = var.dr != "standby" && contains(["dedicated-ui", "advanced"], var.logscale_cluster_type) ? 1 : 0

# Ingest node pool (logscale-ingest-node-pool.tf)
count = var.dr != "standby" && contains(["advanced"], var.logscale_cluster_type) ? 1 : 0

This means:

  • Primary (dr="active" or dr=""): UI and Ingest node pools are created based on logscale_cluster_type

  • Standby (dr="standby"): UI and Ingest node pools are never created, regardless of logscale_cluster_type

  • After promotion to dr="active": UI and Ingest node pools are created automatically by Terraform

Why UI and Ingest Are Not Created on Standby

The standby cluster intentionally excludes UI and Ingest node pools at the infrastructure level for several reasons:

  • Cost optimization: Azure VM costs are eliminated for UI and Ingest node pools until failover

  • Minimal footprint: Single digest pod handles all functions during initial failover

  • Automatic scale-up: Node pools are created during promotion via terraform apply, not maintained idle

  • Resource efficiency: No idle VMs consuming compute resources or incurring costs

  • Consistent with OCI: Matches the DR implementation pattern used in the OCI LogScale deployment

Node Pool Creation During Promotion

When promoting a standby cluster to active (dr="standby" → dr="active"), Terraform automatically:

  • Creates the UI node pool (if logscale_cluster_type includes "dedicated-ui" or "advanced")

  • Creates the Ingest node pool (if logscale_cluster_type is "advanced")

  • Creates associated NSGs and security rules for each new node pool

  • Associates the new subnets with their respective NSGs

Expected time: Node pool creation typically takes 5-10 minutes per pool, depending on Azure region and VM availability.

Note

The HumioCluster nodeCount and pod scheduling will wait for the node pools to become ready before scheduling UI and Ingest pods.