Grading Rubric

Criterion	Strong (3)	Adequate (2)	Weak (1)
Identified misleading symptom	Quickly identified the cache manifest referencing a missing blob; investigated the registry	Tested --no-cache and confirmed the cache was the issue but took time to find the missing blob	Spent time investigating Dockerfile changes, CI runner config, or Docker version
Found root cause in linux_ops domain	Traced to disk-full condition causing partial blob write; found GC failures	Found the full disk but not the partial write corruption	Assumed it was a registry software bug or network issue during push
Remediated in kubernetes_ops domain	Resized PVC, ran GC, rebuilt cache; set up daily GC schedule	Fixed the immediate issue but did not resize or set up ongoing GC	Switched to --no-cache permanently (workaround, not fix)
Cross-domain thinking	Explained the full chain: disk full -> GC failure -> partial write -> corrupt manifest -> CI failure	Acknowledged the disk/registry connection but missed the CI impact chain	Treated it as a CI configuration problem

Prerequisite Topic Packs

• cicd — needed for Domain A investigation (CI pipeline config, Docker build caching, BuildKit)
• docker — needed for Domain A (image manifests, layer digests, BuildKit cache)
• disk-and-storage-ops — needed for Domain B root cause (filesystem capacity, partial writes)
• k8s-storage — needed for Domain C remediation (PVC resize, persistent volumes)
• container-images — needed for understanding registry garbage collection and retention

Pages that link here

• Container Images
• Disk & Storage Ops
• Docker
• K8S Storage
• Symptoms: CI Pipeline Fails, Docker Layer Cache Corruption, Fix Is Registry GC