ChaosProbe C4: node-drain dose-response (design-corrected re-analysis) — raw experiment archive

Raw, tamper-evident run archive for campaign C4, the exploratory design-corrected re-analysis of the availability axis in the MSc thesis Measuring Placement-Sensitive Resilience under Chaos: A Pre-Registered, Layered Study in Kubernetes (Yvo Hu, University of Amsterdam, 2026). C4 is exploratory and outside the frozen confirmatory Holm family. It recomputes the three construction-limited availability-side results (H3 trough depth, H4 frontier, H5 availability sub-score) under node-drain, whose blast radius is placement-dependent. Its corrected criteria were pre-declared before the new data were examined. 8 complete-block sessions, five cross-node fraction levels f ∈ {0, 0.25, 0.5, 0.75, 1.0}, r=1, on Google Online Boutique (eight 2-vCPU/4-GiB workers, Kubernetes v1.28.6, kube-proxy ipvs). All 8 sessions pass doctor --strict. The frozen pre-analysis SHA-256 manifest (c4-nodedrain-manifest.sha256, 56/56 files) is included. Code, manuscript, and analysis scripts: github.com/yhu02/uva-msc-thesis. License: CC-BY-4.0.