dt-obs-kubernetes

The dt-obs-kubernetes skill provides expert guidance for monitoring and analyzing Kubernetes infrastructure using Dynatrace Query Language (DQL). It solves the problem of investigating cluster, pod, node, and workload health at scale, covering resource optimization, pod failures such as OOMKills and evictions, crash loops, scheduling and placement, cost right-sizing, and security posture, as well as operational event history like pod restarts and OOM events.

The skill is a knowledge base of read-only DQL query patterns organized from core monitoring (cluster inventory, node, pod, and workload monitoring) into advanced topics (configuration analysis via labels and annotations, scheduling and placement, cost optimization, and security and compliance). It documents the relevant Dynatrace entity types (clusters, namespaces, nodes, pods, workloads such as deployments, statefulsets, daemonsets, jobs, cronjobs, and HPAs, plus configuration entities like services, configmaps, secrets, and ingresses), the three main query forms (smartscapeNodes for entities, timeseries for metrics, and fetch logs for log events), core fields, and available CPU, memory, operations, and node metrics. It also clarifies entity disambiguation, such as K8S_POD versus CONTAINER, and points to sibling skills for adjacent domains. Reference files drill into cluster inventory, pod and node placement, ingress, persistent volumes, labels and annotations, workload health, pod debugging, and network policies.

Intended users are SREs, platform engineers, and DevOps practitioners using Dynatrace who need to query Kubernetes health, optimize resources, or audit security posture. The skill is a benign, read-only analytics reference: it composes DQL queries to inspect and summarize telemetry and contains no destructive operations, credential handling, or code execution against clusters. It even scopes itself, directing unrelated questions such as AWS-specific queries or distributed tracing to other skills.