Harnessing the Potential of Prometheus Agent Mode
summary
Empirical tests with Prometheus Agent implemented in Prometheus Operator show clear trade-offs: DaemonSet for node-local scraping and automatic scaling cuts resource use versus an HA server and handles node metrics without node-exporter, but it accepts 4–5s metric loss on node failures and adds scheduling/VPA tuning complexity; StatefulSet for per-cluster agents with remote_write gives stronger reliability and easy multi-tenancy and sharding via relabeling, but needs extra capacity and manual horizontal scaling.
AI-written summary · about
chapters (7)
- 0:00Introduction and context
- 2:08Case study of Eva the farmer
- 4:37Agent versus server
- 6:07DaemonSet mode for dynamic farms
- 14:52StatefulSet mode for stable farms
- 17:36Summary and conclusions
- 19:20Question and answer session
related talks
28mPrometheus Operators in the Wild
31mDeep Dive into the Prometheus Operator Ecosystem
35mDeclarative Multi-Cluster Monitoring with Prometheus
41mHigh Available + Scalable Prometheus with Thanos in Alibaba
36mShould I use OTel (collectors), or is Prometheus good enough?
31mFederated Prometheus Monitoring at Scale
34mAutoscale your Kubernetes Workload with Prometheus - Frederic Branczyk, CoreOS
37mMonitoring Kubernetes with Prometheus