Skip to content

SYSTEM Cited by 5 sources

Helm

Helm is the Kubernetes package manager (CNCF). A Helm chart is a parameterized bundle of Kubernetes manifests (Deployments, Services, ConfigMaps, Ingresses, etc.) with templating, values, and versioned release tracking. Charts are the standard distribution format for OSS Kubernetes software (e.g. systems/temporal, Prometheus, systems/grafana).

Stub page — expand on future Helm-internals sources.

Seen in

  • sources/2024-08-08-figma-migrated-onto-k8s-in-less-than-12-months — Figma cited lack of Helm support in ECS as a material limitation: internal teams wanted to run OSS software like Temporal (workflow engine), but on ECS they would have had to manually port every service to systems/terraform. Helm-on-EKS was one of the enabling properties for the migration.
  • sources/2026-03-23-aws-generali-malaysia-eks-auto-mode — Generali Malaysia calls out Helm charts as a first-class enterprise-platform operating principle: "they use Helm chart as a standardized deployment mechanism." Listed alongside stateless- only pods, immutable pods, and HPA as the four compound K8s discipline rules that make EKS Auto Mode viable. Helm encodes the other three rules as template defaults — this is Helm in the role of platform-wide packaging standard, not just per-team tool.
  • sources/2026-04-06-aws-unlock-efficient-model-deployment-simplified-inference-operator-setup-on-amazon-sagemaker-hyperpod — canonical wiki instance of Helm being migrated away from as a packaging primitive. The SageMaker HyperPod Inference Operator previously shipped as a Helm chart; 2026-04-06 repackages it as a native EKS add-on and ships an official migration script (helm_to_addon.sh) that auto-discovers the Helm release's config, scales down Helm deployments, installs the add-on with the OVERWRITE flag, and preserves rollback backups. AWS's motivating frame: Helm put the lifecycle burden (IAM role creation, dependency-chart version management, release upgrades) on the customer; the EKS add-on moves that burden to the vendor. See patterns/eks-add-on-as-lifecycle-packaging for the pattern and systems/sagemaker-hyperpod-inference-operator for the migrated operator. Helm survives on this platform for customer-authored charts — the add-on path is specifically the distribution channel AWS gives to AWS-authored operators.
  • sources/2025-05-06-redpanda-a-guide-to-redpanda-on-kubernetes — canonical wiki instance of Helm chart as the simpler sibling to a Kubernetes Operator, and of an operator that initially wrapped a Helm chart internally then unwrapped it. Redpanda ships both a Helm chart (for teams wanting simple template-driven deployments) and the production-grade Redpanda Operator (for managed upgrades + dynamic configuration + lifecycle automation). Five-axis comparison canonicalised: managed upgrades
  • rollback, dynamic configuration (CRDs vs Helm-values-redeploy), health checks + metrics, lifecycle automation, multi-tenancy management. The customer-facing Redpanda Operator wrapped the Redpanda Helm chart internally via FluxCD — canonical wiki instance of the bundled-GitOps- dependency anti-pattern. The v25.1.x operator removed both the Flux dep and the Helm-chart wrapping; the standalone Helm chart remains available as the simpler deployment path.
  • sources/2025-12-02-redpanda-operationalize-redpanda-connect-with-gitopscanonical Helm + Kustomize + Argo CD composition wiki instance. 2025-12-02 Redpanda tutorial pins the Redpanda Connect Helm chart at charts.redpanda.com/connect targetRevision: 3.1.0. Two canonical deployment shapes: (a) Argo CD multi-source Application — Helm source + separate values repo referenced via $values/... — for Standalone mode; (b) Kustomize wraps HelmhelmCharts in kustomization.yaml + configMapGenerator for pipeline ConfigMaps — for Streams mode. Kustomize's Helm inflation needs --enable-helm build option gated behind supply-chain concerns. Canonical instance of Helm chart as the substrate other composition layers (Kustomize, Argo CD values-split) wrap rather than replace.
Last updated · 542 distilled / 1,571 read