Add K8s GPU metrics collection (CW, DCGM, Prometheus) - #21

Merged

maksimov merged 13 commits into

feature/k8s-gpu-metrics

Apr 19, 2026

Commits on Apr 19, 2026

Add K8s GPU metrics collection design spec
maksimov
committed
22cf265
Add K8s GPU metrics collection implementation plan
maksimov
committed
ee8e309
Add EnrichK8sGPUMetrics for CloudWatch Container Insights GPU metrics
maksimov
committed
879f2c1
Add ProxyGet to K8sClient interface for pod API proxy
maksimov
committed
9a176fa
Add DCGM exporter scraping for K8s GPU metrics
maksimov
committed
4f54360
Add Prometheus query enrichment for K8s GPU metrics
maksimov
committed
98003ef
Add ruleK8sLowGPUUtil for utilization-based K8s GPU waste detection
maksimov
committed
1a35f95
Wire K8s GPU metrics fallback chain into CLI scan flow
maksimov
committed
89d9cb3
Fix DCGM node matching and CW error spam
maksimov
committed
c4dff65
Fix DCGM scrape spam and Prometheus node name mismatch
maksimov
committed
d89df5f
Include time window in low GPU utilization recommendation text
maksimov
committed
fa00dff
Skip CW enrichment when AWS creds unavailable, reduce DCGM noise
maksimov
committed
51db9f4
Merge origin/master into feature/k8s-gpu-metrics
maksimov
committed
0a1b2a1