Argo at Scale: Navigating Complex Multi-Dimensional Deployments Across H...
summary
Adobe runs 430 clusters across 7 providers in 22 regions and deploys at scale — 554 addon deployments in February to a Fleet of ~50,000 Argo CD applications. The team uses Cluster API, ACK, Argo CD/ApplicationSet, Argo Workflows, Prometheus/Grafana and kube-state-metrics to emit custom Prometheus metrics from Argo CD app CRs (sync start/end, sync windows), join via Prometheus rewrite rules, filter false alerts (deny windows, long-running syncs) and route enriched Slack alerts to the committer. A reproducible demo uses vcluster and an IDP Builder CLI and proves the approach works.
AI-written summary · about
chapters (9)
- 0:00Argo at Scale intro
- 2:55Cluster grouping with ApplicationSets
- 6:56Argo CD sync failures
- 9:25Custom metrics for Argo CD
- 12:57Filtering alerts in large fleets
- 13:19Demo setup with IDP Builder CLI
- 16:08Sync windows for alerts
- 19:40Enriching alerts with commit data
- 21:26Future plans for upstream
related talks
34mInfrastructure at Scale Using ArgoCD
36mHow Adobe Planned For Scale With Argo CD, Cluster API, And VCluster
18mUnveil the Secret Ingredients for Argo CD in the Enterprise-Scale
15mA Quantitative Study on Argo Scalability
25mSailing Across the Ocean of Argo Ecosystem Like a Pro
30mArgo’s Vibrant Ecosystem and Community
26mManaging GitOps Deployments in Multi-cluster Production Environments
27mExploring Argo