Skip to main content

Pod Resource Scanner

Kubernetes

Pod Resource Scanner is a read-only Kubernetes monitoring tool that runs as a CronJob. It collects CPU, memory, and disk usage across every namespace and node, writes the results to an append-only CSV, and optionally pushes a formatted dashboard to Google Sheets with scaling recommendations.

Repository: clouddrove/pod-resource-scanner

Supported clusters: EKS, AKS, GKE, and any Kubernetes 1.21+ cluster.

Features​

  • Read-only — only lists pods, nodes, namespaces, and workloads; never modifies the cluster
  • Append-only CSV — single all-resources.csv file with scan timestamps for historical tracking
  • Scaling recommendations — suggests node additions, consolidations, or limit adjustments based on thresholds
  • Week-over-week comparison — tracks namespace resource changes and fires growth alerts
  • Google Sheets dashboard — optional export with timestamped run tabs and a live KPI dashboard tab
  • Cluster-agnostic — works on any standard Kubernetes distribution

What it collects​

CategoryData points
PodsNamespace, node, workload type, resource requests/limits, status
NodesCapacity, allocatable CPU/memory/storage, utilization %
NamespacesPod/container counts, total requested resources, growth %
RecommendationsScale up, scale down, or adjust limits per node/namespace

Installation​

helm install pod-resource-scanner ./chart \
  --namespace pod-resource-scanner \
  --create-namespace \
  --set fullnameOverride=pod-resource-scanner \
  --set image.repository=ghcr.io/clouddrove/pod-resource-scanner \
  --set image.tag=latest

The chart creates a ClusterRole and ClusterRoleBinding with read-only access to nodes, namespaces, pods, and workloads.

Trigger a manual scan​

kubectl create job \
  --from=cronjob/pod-resource-scanner \
  manual-$(date +%s) \
  -n pod-resource-scanner

kubectl logs -n pod-resource-scanner job/manual-<timestamp> -f

Local development (no cluster required)​

pip install -r requirements.txt
export POD_SCANNER_OUTPUT_DIR=./output
python scanner.py

Configuration​

Environment variableDefaultDescription
POD_SCANNER_OUTPUT_DIR/outputDirectory to write all-resources.csv
POD_SCANNER_CLUSTER_NAME(empty)Label added to each row for multi-cluster setups
POD_SCANNER_UPDATE_GOOGLE_SHEET(unset)Set to true to enable Google Sheets export
POD_SCANNER_UTIL_SCALE_UP_PCT75Node utilization % that triggers a scale-up recommendation
POD_SCANNER_UTIL_SCALE_DOWN_PCT25Node utilization % that triggers a scale-down recommendation
POD_SCANNER_GROWTH_ALERT_PCT20Week-over-week namespace growth % that fires an alert
POD_SCANNER_LOG_LEVELINFOLogging verbosity (DEBUG, INFO, WARNING)

Google Sheets export (optional)​

The scanner can write a formatted spreadsheet with one tab per run and a live dashboard tab.

Setup steps​

  1. Enable the Google Sheets API in your GCP project
  2. Create a Service Account and download its JSON key
  3. Create a new Google Sheet and share it with the service account email (Editor)
  4. Create a Kubernetes secret with the Sheet ID and credentials:
kubectl create secret generic google-sheets-creds \
  --from-literal=sheet-id=YOUR_SHEET_ID \
  --from-file=credentials.json=./sa-key.json \
  -n pod-resource-scanner
  1. Enable Google Sheets in Helm:
# values.yaml
env:
  POD_SCANNER_UPDATE_GOOGLE_SHEET: "true"
googleSheets:
  enabled: true
  secretName: google-sheets-creds

Testing​

Run the full test suite without a live cluster:

pip install -r requirements.txt
python3 -m pytest tests/ -v

Preview the spreadsheet layout locally:

python scripts/generate_sample_excel.py

Production checklist​

  • Use a pinned image tag instead of :latest
  • Set POD_SCANNER_CLUSTER_NAME if running across multiple clusters
  • Adjust UTIL_SCALE_UP_PCT and UTIL_SCALE_DOWN_PCT to match your team's thresholds
  • Monitor CronJob success status and the freshness of all-resources.csv
  • Rotate service account credentials regularly if using Google Sheets