Intermittent Kubernetes DNS failures traced to one broken CoreDNS pod, not CoreDNS itself
A WordPress site hosted on a Kubernetes platform began intermittently failing to connect to its MySQL database, with roughly one in three page loads returning an error. Initial investigation ruled out MySQL, which showed no connection saturation or slow queries and was responding normally to direct connections. The real clue came from Redis logs in the same pod, which explicitly reported DNS resolution failures, pointing to a DNS-layer issue rather than a database problem. Further analysis revealed that one of three CoreDNS endpoints was silently dropping all DNS queries routed to it, causing approximately 33% of lookups to fail due to round-robin load balancing across the three pods. The fault was isolated by querying each CoreDNS pod IP individually from a debug pod pinned to the affected node, rather than relying on the ClusterIP service that masked which backend was responsible.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in