DNS Failure Observability and Diagnosis in Kubernetes

Explore DNS failure observability and diagnosis in Kubernetes clusters through this informative conference talk. Delve into common DNS lookup failure scenarios and learn about CoreDNS's built-in observability plugins. Discover novel approaches for monitoring and diagnosing DNS issues, including the use of DNSTAP protocol and eBPF-based tools. Gain insights into diagnostic procedures and best practices for addressing DNS failures in large-scale Kubernetes environments. Understand the various types of DNS resolution problems, their root causes, and effective methods for troubleshooting and resolving them in both intermittent and continuous failure scenarios.

Intro
DNS in Kubernetes
Under the hood
Surprises
DNS becomes fragile
Resource limits
Conflicts
Bugs
Built-In Observabilities
Loggings - log
Loggings - dump
Loggings - debug
Loggings - dnstap - how to analyze
Tracing - trace
Metrics - prometheus
BPF-based Tools for diagnosis
Diagnostic Procedure
Conclusion