Fleeting Metrics: Monitoring Short-lived or Serverless Jobs with Prometheus
Talkat PrometheusDay Europe 2022 · on Tuesday, 2022-05-17· at 13:30· Pavilion 3, Room F | Level 2 | Central Forum + Live Stream, Feria Valencia, Spain + Live Stream·
summary
Defines the problem: Prometheus misses short-lived jobs and pull scraping yields gaps. Lays out three concrete approaches and trade-offs: event-based pipelines that derive metrics from traces/logs using OpenTelemetry, Loki or Tempo (high ingestion cost, sampling loses accuracy); push/partial-aggregation via Pushgateway or aggregation gateways (staleness, cardinality, single-point scaling limits; mentions a Rust aggregating gateway); and open/closed-box heuristics using kube-state-metrics, proxies or eBPF as low-cost, reliable alternatives. Demonstration shows patterns and pitfalls rather than a single silver-bullet solution.
AI-written summary · about
chapters (5)
- 0:00Introduction to fleeting jobs
- 3:56Monitoring challenges
- 11:25Event-based observability
- 19:10Push-based metric solutions
- 26:44Alternative strategies
related talks
25mMonitoring in a Serverless World
27mObservability with Prometheus and beyond
29m“If you Don’t Monitor your Infrastructure, you Don’t Own it!” Regain Control Thanks to Prometheus
36mObservability with Prometheus and Beyond: Leveraging Cloud-native Technologies Outside of Cloud
32mDemocratizing Application Monitoring with Prometheus
24mMonitoring Microservices The Right Way
11mLightning Talk: Searching for Serverless Candidates in the Jungle of Microservices
21mMonitoring Serverless Workloads with OpenTelemetry and Prometheus