The Fluent Bit team is excited to announce the general availability of two new Helm charts, designed to cater to specific use-cases in log aggregation and collection. The Fluent Bit Aggregator Helm Chart and the Fluent Bit Collector Helm Chart are now available on Artifact Hub, providing users with streamlined deployment options for their logging infrastructure.

These charts originated in the stevehipwell/helm-charts repository, where they have been developed and used for the past 5 years. They have now been adopted into the official fluent/helm-charts repository, making them part of the core Fluent Bit project. The existing fluent-bit Helm chart remains available, but users are encouraged to migrate to the new charts for a better experience.

We encourage users to explore these new charts and take advantage of the enhanced capabilities they offer for log management in Kubernetes environments.

Why New Charts?

The original Fluent Bit Helm chart is general-purpose, but running Fluent Bit as a log collector on each node is a fundamentally different use-case from running it as a centralized log aggregator. Each use-case has different requirements for workload type, scaling, storage, networking, and security. By splitting these into dedicated charts, each chart can provide sensible defaults, purpose-built configuration, and a simpler user experience for its specific use-case.

Fluent Bit Collector

The Fluent Bit Collector Helm Chart deploys Fluent Bit as a DaemonSet, running a pod on every node in the cluster to collect logs and forward them to a backend.

Key features include:

  • Kubernetes log collection out-of-the-box — pre-configured to tail container logs from /var/log/containers/ with CRI multiline parsing and Kubernetes metadata enrichment via the kubernetes filter processor.
  • Kubernetes tag-based routing — logs are tagged with kube.<namespace>.<pod>.<container>, making it easy to route logs based on namespace, pod, or container name.
  • Filesystem-backed buffering — configurable host filesystem storage for data durability to protect against data loss during restarts or backpressure.
  • YAML-based pipeline configuration — the chart uses Fluent Bit’s native YAML configuration format, making it easier to read and maintain compared to classic INI-style configuration.
  • Hot-reload support — an optional sidecar container watches for configuration changes and triggers a Fluent Bit hot-reload, enabling configuration updates without pod restarts.
  • Prometheus monitoring — built-in support for creating ServiceMonitor or PodMonitor resources for the Prometheus Operator.

Fluent Bit Aggregator

The Fluent Bit Aggregator Helm Chart deploys Fluent Bit as a StatefulSet, designed to receive logs from multiple collectors (or other sources) and forward them to a central destination. This pattern is useful for centralizing log processing, applying cross-source transformations, and managing output connections to external systems.

Key features include:

  • StatefulSet with headless service — provides stable network identities and ordered deployment, which is important for reliable log aggregation.
  • Forward input by default — pre-configured with a forward input plugin to receive logs from Fluent Bit collectors or Fluentd forwarders.
  • Persistent storage — optional PersistentVolumeClaim support for filesystem-backed buffering, protecting against data loss when aggregator pods are restarted.
  • Horizontal Pod Autoscaling — built-in HorizontalPodAutoscaler support to scale the aggregator based on resource utilisation or custom metrics.
  • Pod Disruption Budgets — configurable PodDisruptionBudget to ensure availability during cluster maintenance.
  • Ingress support — configurable ingresses for each input plugin, allowing external sources to forward logs into the aggregator.
  • Upstream servers — support for defining upstream server groups for round-robin data distribution across output backends.
  • Grafana dashboards — optional Grafana dashboard installation for monitoring aggregator performance, including input/output byte and record rates, error rates, and resource utilisation.

Shared Features

Both charts share a number of common capabilities:

  • YAML-based configuration — pipeline configuration is defined in Helm values using Fluent Bit’s native YAML format, eliminating the need to manage raw configuration files.
  • Hot-reload — optional sidecar-based hot-reload support for applying configuration changes without restarting pods.
  • Lua scripting — Lua scripts can be defined in values and are automatically mounted for use in filter plugins.
  • OCI and non-OCI installation — charts are published as OCI artifacts to ghcr.io/fluent/helm-charts (recommended) and via the traditional Helm repository at https://fluent.github.io/helm-charts/.
  • Cosign verification — OCI chart releases are signed using Cosign, allowing users to verify chart integrity before installation.
  • Prometheus Operator integration — built-in ServiceMonitor support for exposing Fluent Bit metrics.
  • Security-first defaults — non-root containers, read-only root filesystems, and dropped capabilities out-of-the-box.

Installation

1
2
3
4
5

# Aggregator
helm upgrade --install fluent-bit-aggregator oci://ghcr.io/fluent/helm-charts/fluent-bit-aggregator

# Collector
helm upgrade --install fluent-bit-collector oci://ghcr.io/fluent/helm-charts/fluent-bit-collector

Non-OCI Repository

1
2
3
4
5
6
7
8

helm repo add fluent https://fluent.github.io/helm-charts/
helm repo update

# Aggregator
helm upgrade --install fluent-bit-aggregator fluent/fluent-bit-aggregator

# Collector
helm upgrade --install fluent-bit-collector fluent/fluent-bit-collector

Repository Improvements

As part of introducing these new charts, we’ve also made significant improvements to the fluent/helm-charts repository:

  • Automated release pipeline — charts are now released automatically via a reusable workflow when the chart version is updated, with support for OCI publishing, GitHub Pages, and Artifact Hub linting.
  • PR validation — pull requests are validated with helm-docs generation, changelog enforcement, Artifact Hub linting, and Kubeconform schema validation against the latest Kubernetes versions.
  • OpenSSF Scorecard — the repository now runs the OpenSSF Scorecard to track security best practices.

Getting Started

Visit the chart pages on Artifact Hub for full documentation and default values:

If you have questions or feedback, please open an issue on the fluent/helm-charts repository or join the conversation on the Fluent Slack.