Kubernetes Operations

Kubernetes has the deepest Cloud Ops integration — pod management, real-time log streaming, interactive container shells, resource browsing with visual graphs, inline YAML editing, and resource deletion. All operations work through both the web console and CLI, in both developer mode and admin mode.

Viewing Pods

Pod viewing is the starting point for most Kubernetes operations. You can find pods by their managing resource (Deployment, StatefulSet, DaemonSet, Job, CronJob) and see their current state at a glance.

In the Web Console

Navigate to any Kubernetes cloud resource and open the Kubernetes tab. The pod list shows each pod with its status, ready state, restart count, and an actions menu. Click a pod name to view its full details.

From the actions menu on any pod, you can:

• Stream Logs — Open the real-time log viewer
• Exec into Container — Open a browser-based terminal
• Delete — Remove the pod (with confirmation)

Using the CLI

Find pods for a cloud resource (developer mode):

planton kubectl get pods -r my-org/prod/KubernetesDeployment/payments-api

Find pods using a direct connection (admin mode):

planton kubectl get pods --connection k8s-prod-cluster -n payments

The -r flag accepts a cloud resource reference in the format org/env/Kind/slug. The --connection flag accepts a provider connection slug and requires -n to specify the namespace.

Streaming Logs

Log streaming delivers real-time log output from running pods, with filtering to narrow down to exactly what you need.

In the Web Console

Click "Stream Logs" on any pod to open the log viewer. Logs begin streaming immediately and appear in a scrollable, read-only editor.

The log viewer provides:

• Play/Pause — Pause log updates to read through output, then resume
• Filters — Expand the filter panel to narrow logs by container name, time range, tail lines, or content search
• Copy — Copy all displayed logs to the clipboard
• Clear — Clear the displayed log buffer
• Restart — Stop and restart the log stream (useful after changing filters)

Each log line includes the pod name and container name as a prefix, so you can distinguish output when streaming from pods with multiple containers.

Using the CLI

Stream logs for pods managed by a deployment (developer mode):

planton kubectl logs deployment payments-api -r my-org/prod/KubernetesDeployment/payments-api

Stream logs with filters (admin mode):

planton kubectl logs deployment payments-api --connection k8s-prod -n payments \
  --tail-lines 500 \
  --since-duration 30m \
  --container-name-filter app \
  --content-filter "ERROR"

The logs command takes two positional arguments: the pod manager kind (e.g., deployment, statefulset, pod) and the name. Available flags:

Exec into Containers

Container exec gives you an interactive shell inside a running container — directly from the browser or CLI. No kubectl required, no kubeconfig to manage.

In the Web Console

Click "Exec into Container" on any pod. A terminal drawer opens at the bottom of the screen.

The workflow:

1. Select a container — If the pod has multiple containers, pick the one you want to shell into
2. Choose a shell — Type the shell name when prompted (bash, sh, or another available shell)
3. Run commands — The terminal connects and you get a live shell session

The terminal supports standard input, output display, and scrolling. You can expand the drawer to full height for more space. Type exit or click the close button to end the session.

Common use cases:

# Check environment variables
env | grep DATABASE

# Verify file mounts
ls -la /app/config

# Test internal connectivity
curl http://other-service:8080/health

# Inspect application state
cat /tmp/debug.log

Using the CLI

Exec into a container (developer mode):

planton kubectl exec my-pod-7d8f9c6b4d-x2k9m app -r my-org/prod/KubernetesDeployment/payments-api

Exec with a specific shell (admin mode):

planton kubectl exec my-pod-7d8f9c6b4d-x2k9m app --connection k8s-prod -n payments --shell bash

The exec command takes two required positional arguments: the pod name and the container name. You can optionally append a command to run instead of an interactive shell:

planton kubectl exec my-pod app -r my-org/prod/KubernetesDeployment/payments-api -- cat /etc/config/app.yaml

Browsing Resources

Beyond pods, Cloud Ops lets you browse, inspect, edit, and delete any Kubernetes resource in a namespace.

Listing Resources

In the web console, the Kubernetes tab on a cloud resource shows all resources organized by kind. The namespace graph view displays resources as a directed acyclic graph (DAG), visualizing relationships between Deployments, ReplicaSets, Pods, Services, ConfigMaps, and other resources.

Using the CLI:

# List all resources in the namespace
planton kubectl get all -r my-org/prod/KubernetesDeployment/payments-api

# List resources by kind
planton kubectl get deployments -r my-org/prod/KubernetesDeployment/payments-api

# List namespaces (admin mode only)
planton kubectl get namespaces --connection k8s-prod

The get command supports multiple output formats:

Inspecting Resources

View the full details of any resource — equivalent to kubectl describe plus the complete YAML definition.

In the web console, click any resource node in the graph or any resource in a list to see its describe output and YAML.

Using the CLI:

# Describe a resource
planton kubectl describe deployment payments-api -r my-org/prod/KubernetesDeployment/payments-api

# Get a specific resource
planton kubectl get deployment payments-api -r my-org/prod/KubernetesDeployment/payments-api

Editing Resources

Edit a resource's YAML directly. In the web console, click "Edit" on any resource to modify its YAML in the browser and apply changes immediately.

Using the CLI, the edit command opens the resource YAML in your $EDITOR (or vi), and applies changes when you save and close:

planton kubectl edit deployment payments-api -r my-org/prod/KubernetesDeployment/payments-api

Common edits include adjusting resource limits, updating environment variables, modifying labels or annotations, and changing replica counts.

Deleting Resources

Delete a resource with confirmation. In the web console, click "Delete" on any resource — a confirmation dialog appears before the deletion proceeds.

Using the CLI:

planton kubectl delete pod payments-api-7d8f9c6b4d-x2k9m -r my-org/prod/KubernetesDeployment/payments-api

CLI Reference

All planton kubectl subcommands share three persistent flags that control the access mode:

Command Summary

Best Practices

Use the Web Console for Quick Debugging

The browser-based terminal and log viewer are the fastest path to understanding what's happening in a running pod. No local tooling setup required — open the cloud resource, click a pod, and start debugging.

Use the CLI for Automation and Scripting

The CLI supports the same operations with structured output formats (table, yaml). Use it in scripts, CI pipelines, or when you prefer terminal-based workflows.

Prefer Developer Mode for Service Work

Developer mode scopes operations to your deployment's namespace automatically. You cannot accidentally inspect or modify resources belonging to other deployments. This is the safer default for day-to-day service work.

Use Admin Mode for Cluster-Wide Operations

Admin mode gives you access to any namespace accessible through the connection. Use it for cross-cutting operational tasks like inspecting kube-system, checking node-level resources, or troubleshooting cluster infrastructure.

Remember That Manual Edits Are Temporary

Changes made through Cloud Ops (editing a Deployment's replica count, modifying a ConfigMap) take effect immediately but may be overwritten by the next deployment. For permanent changes, update your infrastructure code or service configuration and redeploy.

Related Documentation

• Operations Overview — What Cloud Ops is, dual access modes, supported providers
• Resource Browser — Multi-cloud resource listing for AWS, GCP, and Azure
• Runner — The secure execution agent that processes Cloud Ops requests
• Connections > Kubernetes Clusters — How Kubernetes cluster credentials are managed