Integrating Fluent Bit Into a PaaS
summary
DigitalOcean App Platform uses Fluent Bit as a daemonset and a per-cluster operator to collect logs into a central system via a custom websocket plugin and to route customer logs to syslog/HTTP destinations. Container sandboxing with gVisor breaks tail input so the stat watcher option added in Fluent Bit 1.8 becomes essential. The operator initially lacks a Go client and causes high CPU while rendering a single giant config secret; Prometheus metrics are truncated to 32-character plugin names, so the platform generates recording rules and considers gzip for secrets.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:13Agenda
- 0:50Overview
- 2:10Architecture
- 3:20Internal Logging
- 4:07Application Logging
- 6:45Data Pipeline
- 7:21Input
- 8:34Modify Filter
- 8:47Output
- 9:15Monitoring
- 11:09Issues
- 12:27Future work
related talks
22mLogging: Fluentd & Fluent Bit
22mFluent Bit - Swiss Army Tool of Observability Data Ingestion
27mBuild a Cloud Native Logging Pipeline on the Edge with Fluentbit Operator
36mDeep Dive: Fluent Bit
34mPower of O11y with FLUENT BIT in AWS EKS environment
40mFluentd Deep Dive
12mLightning Talk: Achieving Unified Observability for Cloud and Edge with FluentBit
13mFluent Bit