Infrastructure
The Infrastructure and Kubernetes sections show the health of what runs underneath your services — from a cluster-wide summary down to a single pod.
Overview
Section titled “Overview”The Overview page is the at-a-glance picture: how many nodes are up, how many workloads are running, how many are unhealthy, and where your CPU is going — including the top workloads and nodes by usage.
Workloads
Section titled “Workloads”Workloads shows resource usage per workload, with replica health and restart counts so a crash-looping or under-provisioned workload stands out immediately. Open a workload to see its trend, its pods, and jump straight to its Traces.
Kubernetes
Section titled “Kubernetes”The Kubernetes section goes from a whole cluster down to a single pod, and every page links up and down that path.
Clusters
Section titled “Clusters”Clusters lists each Kubernetes cluster with its health, Kubernetes version, nodes ready, pods running, and degraded workloads, plus how much CPU and memory pods have requested against what the nodes can schedule, and what’s actually in use. Open a cluster for its capacity, its nodes, a breakdown by namespace, the pods needing attention (not running, not ready, or restarting), and its recent warnings.
Nodes lists your whole fleet with each node’s status and health, CPU and memory usage, capacity, and when it was last seen. Open a node to see:
- its health — whether it’s ready, and any problem it reports (memory, disk, or process pressure, network unavailable, a kernel or runtime fault), with the reason and how long it’s been going on; whether it’s cordoned, and its taints;
- requested vs allocatable — how much of its CPU, memory and pod capacity is already committed to the pods on it, next to its actual CPU and memory over time;
- the pods running on it, each with status, readiness, restarts and usage, linked to the pod and to the workload that owns it.
Recent warnings
Section titled “Recent warnings”When Kubernetes reports a problem — a pod that can’t be scheduled, a container crash-looping or killed for running out of memory, a failing health probe, an image that won’t pull, a node going not-ready — it shows up under Recent warnings on the cluster, node, workload, and pod it concerns.
Warnings are also ordinary log lines, at warning level, under the workload they
concern, so you can search them on the Logs page (select them with
@ultrvlt.source:k8s-event, or a specific kind with @k8s.event.reason:OOMKilling)
and alert on them with a log alert rule.
Object browser
Section titled “Object browser”You can also browse the objects themselves — Pods, Deployments, DaemonSets, StatefulSets, and ReplicaSets. Each list shows status and readiness; open an object to walk its ownership (from a pod up to its controller and back down), see its labels and containers, and tail container logs right there.