Last updated:
Explore Kubernetes in Dash0Release historyThis feature was introduced by Explore your Kubernetes pods, live or historical on November 16, 2026.
The Kubernetes section under Infrastructure shows your clusters as Kubernetes describes them: pods and the workloads that own them, with the metrics, logs, traces and events they produce, and the manifests behind them.
Requirements
- Telemetry from your clusters: Install the Dash0 operator. It discovers clusters, nodes, workloads and pods, and collects their metrics, logs and events.
- Manifests: Kubernetes object collection, from the operator once you turn it on, or from a collector of your own running the
k8sobjectsreceiver. The YAML tab, a container's last termination, and the no-memory-limit callout all read the collected manifest. See The YAML Tab.
Pods
Infrastructure → Kubernetes → Pods has two views of the same pods, chosen with the View control.
Map
The Map groups pods into their clusters and namespaces and colors each one by health. Click a cluster to zoom into it, a namespace to zoom further, and a pod to open its side panel. Press Esc, or click outside, to zoom back out.
Needs attention lists why pods are unhealthy and how many each reason affects. Each pod counts once, under its most severe reason.
| Reason | Meaning |
|---|---|
| Failed | The pod's phase is Failed. |
| Pending | The pod's phase is Pending. |
| Unknown phase | The pod's phase is Unknown. |
| Not ready | The pod is running, but not every container in it is ready. |
| Near the CPU limit | The pod uses more than 80% of its CPU limit. |
| Near the memory limit | The pod uses more than 80% of its memory limit. |
Table
The Table lists pods as rows with Namespace, Age, Status and Restarts, and the Usage and Limit used of both CPU and memory. An IP column can be added. Group by groups the rows by cluster, namespace or node. The filter bar narrows the list by name, cluster, namespace, node or IP.
Keyboard Shortcuts
The pod table takes the keys k9s uses.
| Key | Action |
|---|---|
j / k | Next / previous pod |
g / Shift+G | First / last pod |
Enter or d | Open the pod's side panel |
y | Open the pod's YAML |
l | Open the pod's logs |
e | Open the pod's events |
o | Open the pod's node |
p | Open the parent workload |
c | Copy the pod name |
/ | Filter pods |
Shift+N | Sort by name |
Shift+A | Sort by age |
Shift+S | Sort by status |
Shift+C | Sort by CPU |
Shift+M | Sort by memory |
The Pod Side Panel
Opening a pod gives six tabs.
Overview
Summary names the pod's owners and the services it runs, each a link.
Containers lists every container with its state, total restarts, CPU and memory. Hovering over a container's state shows its Current state and Last termination, with the Reason, Exit code, Signal, and the times it started and finished. This is where an OOMKilled or a non-zero exit code shows up. These details come from the collected manifest, so they need object collection and show only in the live view.
Restarts puts the pod's restarts on a timeline across the selected range.
CPU usage, Memory usage and Network I/O chart the pod against its requests and limits. When object collection is on, a container with no memory limit is called out, because it can use memory until the node runs out.
Events lists the Kubernetes events for the pod. Kubernetes records no event when a container is killed or fails, so Dash0 adds those rows from the container's status metrics instead. They are marked, and their time is approximate.
Related resources is a tree, scoped to the pod's namespace: the pod's cluster, its nodes and the namespace's pods on them, and the owning workload down through its ReplicaSets to its pods.
Metrics
CPU usage against requests and limits, memory against its limit, container restarts, ephemeral storage, and network errors per second.
Logs
The pod's logs, searchable, and filterable to one container. When a container terminated without writing logs in the range, the tab says so and links to the events that explain it.
Traces
Spans from the pod, in two modes. Spans lists them by time. Operations groups them by operation name with span count, errors, p50 and p95. Opening a span shows its trace without leaving the page; View all in Tracing hands the selection to the Tracing page.
YAML
The pod's manifest as it was last collected. Collection is off by default. See The YAML Tab.
Attributes
Every resource attribute the pod's telemetry carries, including cloud and host attributes.
The YAML Tab
The YAML tab shows the object's manifest as it was last collected, with Observed and the time. An object that has since been deleted still shows its last known state.
Manifests arrive as log records from the OpenTelemetry Collector's k8sobjects receiver. Either the Dash0 operator runs that receiver for you, or you run it in a collector of your own.
With the Dash0 Operator
Turn Kubernetes object collection on in two places, both of which default to false:
spec.kubernetesObjectCollection.enabledon the operator configuration resource, for the cluster.spec.kubernetesObjectCollection.enabledon the Dash0Monitoring resource of each namespace you want manifests from.
With both set, the operator collects the manifests of pods, Deployments, ReplicaSets, StatefulSets and DaemonSets in those namespaces, and of nodes and persistent volumes for the cluster. Its own namespace is never collected. The operator's cluster-metrics-collector keeps these manifests in memory, so check Kubernetes Object Collection for its limits and sizing before you turn it on for a large cluster.
Kubernetes sends an object only when it changes, and the operator does not read the current state of pods and ReplicaSets when collection starts. A pod's manifest therefore arrives the next time that pod changes, not right after you turn collection on.
Before a manifest leaves your cluster, the operator masks the values of environment variables and removes the annotations that kubectl, kapp, Rancher and Banzai Cloud use to store a serialized copy of the whole manifest. Annotations of other tools that embed manifest copies are not recognized and are not removed.
With Your Own Receiver
Run the k8sobjects receiver in a collector that exports logs to Dash0, and the tab shows the objects it sends. Two things to get right:
- Cluster identity: the manifests need the same
k8s.cluster.nameandk8s.cluster.uidresource attributes as the cluster's other telemetry, or they land under a different cluster than the pod you opened. - Redaction: the masking described above is the operator's work, not Dash0's. Remove environment variable values and the annotations holding a copy of the manifest, such as
kubectl.kubernetes.io/last-applied-configuration, in your own pipeline before exporting.
When the tab reports no YAML:
| Message | Cause |
|---|---|
| No YAML collected | Nothing was collected for this object in the last 7 days. Below the message, the tab names the cause when Dash0 can tell: collection is off for the cluster, the namespace is not opted in, the operator is too old to collect manifests, or collection is on but the object has not changed since. |
| Cluster unknown | Objects with this name were collected from more than one cluster, and the pod's telemetry does not say which. Set clusterName in the Dash0OperatorConfiguration resource, or set k8s.cluster.name with a resource processor in your own collector. |
Live and Historical Views
The Kubernetes pages are live: they follow the cluster as it changes. The Live control pauses that, so a row stays still while you read it, and resumes it again.
The same control picks a past window: yesterday, the day before, this day last week, the previous week, or an absolute range. A banner then says you are looking at the cluster as it was, and offers Back to live. The YAML tab is the exception. It always shows the current manifest and warns that it may differ from the one in place at the time.
Further Reading
- About Kubernetes Monitoring - Ways to monitor Kubernetes with Dash0
- Kubernetes Object Collection - Turn on object collection, and size the collector for it
- Explore Resources - Kubernetes resources alongside the rest of your infrastructure
- Kubernetes Metrics - Metrics collected by Dash0