Documentation
Documentation
Drop pre-caches container images on Kubernetes nodes using short-lived Pods.
Why
When many CI jobs or workloads start simultaneously, Kubernetes nodes face a thundering herd of image pulls. Concurrent pods on the same node all pulling the same large image saturate bandwidth, stall containerd, and cascade into failures.
| Problem | Impact |
|---|---|
| Thundering herd | Parallel pulls of the same image destabilize nodes |
| Registry overload | Sudden pull surges hit rate limits or cause outages |
| Cold-start latency | Large images delay workloads that need them immediately |
Drop pre-caches images before workloads need them, paces pulls to stay within safe limits, and automatically discovers which images matter most.
Sections
| Section | What you’ll find |
|---|---|
| Installation | Helm install, prerequisites |
| Usage | CachedImage, CachedImageSet, PullPolicy examples |
| Discovery | Automatic image discovery with DiscoveryPolicy |
| Monitoring | Prometheus metrics, events, status conditions |
| Reference | CRD field reference, errors, metrics, architecture |
| Developing | Build, test, lint, project structure |
| For AI Agents | llms.txt, Markdown API, generation architecture |
Last updated on