Intro: Fluentd
summary
Fluentd separates a small core and hundreds of plugins for inputs, filters, buffers and outputs, so logs flow through an input→filter→buffer→output pipeline; the recommended file buffer writes chunks to disk for durability and survives crashes. Chunk size and flush interval are tunable and directly trade throughput for latency. The daemonset image for Kubernetes tails container files and attaches namespace/pod metadata automatically. A lighter C-based Fluent Bit targets low-CPU/memory edge collectors.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:53What's Fluentd
- 4:13Unified logging layer
- 5:03Design Core
- 6:33Event structure
- 7:57Data pipeline simplified
- 8:32Architecture: Input Plugins
- 9:15Architecture: Fiber Plugins
- 10:02Architecture: Butler Plugins
- 12:13Architecture: Output Plugins
- 12:34Divide & Conquer for rubry
- 13:423rd party output plugins
- 14:25Simple forwarding
- 15:55Multiple destinations
- 16:56Multi-tier Forwarding
- 18:52Resources
- 21:16Docker logging with
- 22:40Data collection with fluent-legger
- 23:53Shared data volume and tailing
- 26:28Container Logging approach summary
- 27:44Fluents and Fluent-bit
- 28:38Container Logging with fluent-bit
related talks
23mExporting your trace telemetry with OpenTelemetry and Fluent Bit
49mContainerd Fundamentals
27mShow Me Your Labels and I’ll Tell You Who You Are
21mKubernetes Under the Magnifying Glass: Observability With OpenSearch
45mWhy Spring Belongs In Your Data Stream (From Edge to Multi Cloud)
35mDancing with Helm, Skaffold and Minikube!
33mSeeing the Forest for the Trees: An Operator’s Guide to Observability of CF with...
22mDo Flink on Web with FLOW