Defining SLOs and SLIs for ArgoCD: A Metrics-Driven Approach To Observability
summary
ArgoCD metrics are trimmed to customer-focused SLIs such as 95% of syncs <20s and production-only sync success rates, with historical baselines used to set targets; reconciliation queue latency repeatedly drives the error budget into breach and appears as the single cause on the dashboard. Queries count 404s as success for pre-existing resources, use pending git requests (5m max) and combine git and k8s API latencies, manifest generation from CI/Helm is measured, Terraform stores SLO defs, and a webhook aggregator is proposed to compute end-to-end deployment time.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:27SLIs and SLOs with examples
- 3:55Limitations of traditional monitoring
- 5:43Advantages of SLOs
- 7:11Metrics to meaningful SLIs
- 9:41Setting SLO targets
- 11:49ArgoCD deployment flow
- 13:54Refining ArgoCD SLIs
- 16:26Visualizing SLOs with dashboards
- 17:31Next steps
- 18:54Key takeaways
related talks
35mFrom Monitoring to Observability: Left Shift your SLOs with Chaos
22mLook Ma, No Metrics: A Quest Towards Extracting SLOs
35mAvoiding Goodhart's Law - Use SLO's as Tools Not Cudgels
31mServerless Observability: a case study of SLOs
33mHow and why we lowered our SLO
30mMetrics for good, not evil
32m9 Things You Should Do When Starting to Use SLOs
37mReal World SLOs and SLIs: A Deep Dive