Intermittent DNS Failures from ndots Search Amplification
warningHigh ndots setting (default 5) causes excessive DNS queries as each name is tried with all search domains before FQDN lookup, leading to DNS cache stampede and intermittent failures during pod scaling events.
Monitor coredns_request_count spikes correlating with pod scaling events. Calculate queries per application request - if ratio exceeds 5:1, ndots amplification is likely. Check for timeout errors during high pod churn periods.
Reduce ndots to 2 in pod dnsConfig for applications making external calls. Use fully qualified domain names with trailing dots in application code (e.g., 'api.example.com.'). Add TTL jitter (10% randomization) in CoreDNS config to prevent synchronized cache expirations. Configure serve_stale to handle stampede scenarios.