Flux + Observability: Featuring Prometheus Operator and Pixie
summary
Flux demonstrates concrete observability plumbing: the Flux notification controller forwards controller events to Slack/GitHub with secrets decrypted via GCP KMS and workload identity; Flux exports Prometheus metrics consumed by kube-prom-stack + Grafana and Alertmanager for rate/latency alerting; a Tailscale subnet router exposes cluster service IPs to a laptop without public ingress; Pixie eBPF agentless traces reveal HTTP call paths. Demo largely works though network hiccups delay some dashboards.
AI-written summary · about
chapters (11)
- 0:00Flux and observability intro
- 1:28Flux in GitOps workflows
- 3:57Why observability matters
- 6:27Flux notification controller
- 8:11Setting up Flux notifications
- 12:31Demo: Flux alerts & reconciliation
- 17:20Prometheus stack for Flux
- 20:59Pixie for eBPF monitoring
- 22:06Pixie dashboard & use cases
- 23:51Alert manager rules
- 24:36Flux dashboards summary
related talks
33mFlux Ecosystem Evolution
14mFlexible, Open and Easy Observability for Developers
38mAutomated & Observable Deployments with Flux + Keptn
27mLightning Talk: GitOps, A Slightly Realistic Situation on Kubernetes with Flux
34mTotal Visibility: Harnessing Pixie for a Full Observability Landscape
39mEnhance Your Kubernetes Experience with GitOps & Flux
20mGitOps Modern Operations for EKS
15mLightning Talk: GitOps and Progressive Delivery with Flagger, Istio and Flux
