1 comment

[ 4.1 ms ] story [ 8.0 ms ] thread
Hi HN — I built kroot to help debug Kubernetes failures by tracing dependencies between resources.

Instead of only showing symptoms (CrashLoopBackOff, etc), it builds a dependency graph of cluster resources and tries to answer:

"Why did this actually break?"

Example failure chain it can detect:

Secret/prod/db-password missing ↓ Pod/prod/payments-api CrashLoopBackOff ↓ Service/prod/payments has no endpoints ↓ Ingress returning 503

Would love feedback from people running Kubernetes in production.