oxygen914

Diagnose Pending Pods and FailedScheduling from allocatable capacity, requests, quota, taints, affinity, and topology evidence.

Diagnose API server and control-plane availability from live health, latency, quorum, and dependency evidence.

Diagnose database latency, pool saturation, lock contention, and dependency pressure without modifying data.

Diagnose unavailable Deployment replicas and rollout stalls without treating every Pod restart as a Deployment failure.

Diagnose node filesystem and inode pressure from usage, growth, mount, and workload evidence without deleting data.

Diagnose DNS, TCP, Service endpoint, packet-loss, and conntrack symptoms using bounded read-only evidence.

Diagnose node-wide memory pressure and kernel OOM while separating cache, workload, limit, swap, and capacity causes.

Diagnose NodeNotReady and node-exporter loss while separating node root cause from derived Pod alerts.

Diagnose node clock drift and file descriptor pressure as operating-system runtime health failures without changing host state.

Diagnose repeated container exits and CrashLoopBackOff from last state, previous logs, events, and revision evidence.

Diagnose container OOMKilled events and distinguish cgroup limit, leak, traffic, and node-memory causes.

Diagnose Redis latency, memory pressure, blocked clients, eviction, and persistence stalls without destructive commands.