Edge Release List

Improves PostgreSQL failover speed with enhanced shutdown logic, adds workflow composition capabilities with template references, and fixes repository working directory issues in CI/CD workflows.

• pg_shutdown_timeout renamed to pg_smart_shutdown_timeout in kube_pg_cluster
• New shutdown/failover logic reduces PostgreSQL downtime from 60-120s to <5s
• Adds templateRef support to wf_spec for Workflow composition

Improves PostgreSQL storage configuration and memory tuning, simplifies Karpenter node pool management, adds support for custom PostgreSQL parameters and schemas, and fixes several bootstrapping and environment issues.

• pg_storage_increase_percent renamed to pg_storage_increase_gb in kube_pg_cluster
• pg_storage_gb renamed to pg_initial_storage_gb in kube_pg_cluster
• Node pool inputs (node_vpc_id, node_subnets, node_security_group_id) moved from kube_karpenter to kube_karpenter_node_pools

Graduates BuildKit to beta status, updates ECR repository configuration for better customization, adds Argo Events components, introduces several helpful CLI tools, and improves ARM support across core components.

• kube_buildkit graduates to Beta status
• ECR repository input format changed to support per-repository configuration
• New kube_argo_sensor and kube_argo_event_source submodules (Alpha)
• All core cluster components now support both amd64 and arm64 nodes

Introduces an alternative Kubernetes scheduler that significantly improves pod bin-packing for 25-33% cost reduction, adds integrated descheduling for underutilized nodes, and fixes Karpenter deployment issues.

• New kube_scheduler for 25-33% node cost reduction via improved bin-packing
• New panfactum_scheduler_enabled input (default false) to opt in to the new scheduler per module
• Automatic descheduling of pods from low-utilization nodes when scheduler is enabled

Adds PgBouncer configuration options, introduces cost-saving opportunities with enhanced_ha_enabled flag, adds OpenCost for Kubernetes cost analysis, and fixes various issues with VPC, Authentik, and pod scheduling.

• New enhanced_ha_enabled input to core modules — disable for ~$50/month savings with minor availability tradeoff
• New kube_open_cost module for Kubernetes cost analysis
• Many new PgBouncer configuration inputs in kube_pg_cluster

Deprecates kube_trust_manager in favor of kube_reflector, adds domain contact type configuration, simplifies EKS reset process with new command, and fixes VPC and CoreDNS-related issues.

• kube_trust_manager deprecated in favor of kube_reflector
• New pf-eks-reset command consolidates manual cluster reset steps

Introduces workload utility submodule for standardized pod specs, disables VPC flow logs by default, improves environment setup with scaffolding script, and fixes several environment and resource-related issues.

• New kube_workload_utility submodule for production-hardened Pod specs
• VPC flow logs now disabled by default in aws_vpc
• New pf-env-scaffold script for environment bootstrapping setup

Upgrades to devenv 1.0, improves Redis sentinel configuration, significantly speeds up devenv rebuilds on macOS, and fixes several issues with Karpenter autoscaling and ExternalDNS.

• Upgraded to devenv 1.0 — .env values can no longer be referenced directly inside .nix files
• macOS devenv rebuilds now 10-15x faster (from 10+ minutes down to ~45 seconds)
• Fixed Karpenter autoscaling issue when extra_tags was provided