I maintain a Kubernetes clusters that pulls a number of images (mostly from Docker Hub, but also some from GCR and GHCR). I am interested in building a dashboard to measure the amount of time it takes for these images to be pulled (to determine how much this is contributing to pod start latencies). Is anyone aware of means to measure this latency using Prometheus / specific exporters?
Asked
Active
Viewed 104 times
1 Answers
2
For Kubernetes >=1.30
Kubelet now exports a histogram, image_pull_duration_seconds, for exactly this case. See kubernetes/kubernetes#121719 for details.
Derek Brown
- 135
- 5
Elad Avikzer
- 46
- 7