Kubernetes

Know why the pod
keeps restarting.

Install one Helm chart and Monocle shows your pods, workloads, jobs, and nodes next to the traces and logs of the services they run. When something breaks, it tells you what happened in plain words.

Get started Read the docs

One helm install

A node agent and a cluster agent

Every 60 seconds

Metrics, state, and events

Plain diagnostics

Crash loops, OOM kills, pull errors

Linked traces

Each span knows its pod

The state of your cluster, in one page

The overview answers the first questions: is there room left, is everything running, and what needs a look. Switch to This environment to keep only the namespaces your services run in.

CPU and memory used, and reserved by requests, against the capacity of your nodes
Pods and nodes ready, and the workloads that need attention first
The warnings and rollouts of the last hour, from Kubernetes events

Problems explained, not just listed

A pod in CrashLoopBackOff tells you little. Monocle reads the restarts, exit codes, and events, and writes down what happened, with the last log lines before the crash.

Crash loops, OOM kills, images that cannot be pulled, and pods that cannot be scheduled
Failed jobs, missing replicas, and nodes low on memory or disk
Each diagnosis says how often it happened, like "Keeps crashing: restarted 11 times in the last hour"

Every resource, one click away

Pods, containers, workloads, jobs, nodes, namespaces, volumes, and clusters each get their own list, with stat cards on top. Open any of them in a side panel without losing your place.

Search, sort, and group by namespace, node, workload, or status
Side panels with Overview, Metrics, Logs, Traces, and Events tabs
CPU, memory, network, filesystem, and volume metrics from the kubelet

One command to install

The chart runs an agent on each node for kubelet metrics and container logs, and one cluster agent for the state of your resources and the Kubernetes events. Metrics and logs need no change in your apps.

Container logs are read from each node and linked to their pod
Pods, workloads, jobs, CronJobs, and nodes are reported every 60 seconds
Built on the OpenTelemetry Collector, configured through the chart values

terminal

helm install monocle-agent oci://ghcr.io/monocle-sh/charts/k8s-agent \
  -n monocle --create-namespace \
  --set monocle.apiKey=<ingestion key> \
  --set monocle.environment=production \
  --set monocle.clusterName=prod-eu

Traces that know their pod

Point the OTLP exporter of your services at the agent on their node. It adds the pod, workload, namespace, and node to every span, then forwards it with your service’s own API key.

The panel of a pod or workload lists its traces and links to the services it runs
This environment keeps only the namespaces your services run in
Works with the AdonisJS agent and any OpenTelemetry SDK

deployment.yaml

env:
  - name: HOST_IP
    valueFrom:
      fieldRef:
        fieldPath: status.hostIP
  - name: OTEL_EXPORTER_OTLP_ENDPOINT
    value: http://$(HOST_IP):4318

Works with

Plugs into your stack

Helm

oci://ghcr.io/monocle-sh/charts/k8s-agent

Kubernetes

Kubelet metrics, cluster state, and events

OpenTelemetry Collector

The agents run the upstream collector

Container logs

Read on each node, linked to their pod

AdonisJS

Traces tagged with their pod

Any OpenTelemetry SDK

Send OTLP to the agent on the node

Next up

From the click to the database query.

The browser SDK reports errors, logs, and network calls from your web app. Requests to your API carry a trace header, so a click and the backend work it triggered land in the same trace.

Explore Frontend monitoring

Read the docs

All features

Early access

See what your apps
are really doing.

Traces, logs, errors and metrics for every app you ship. Bring it all into focus.