homelab/cluster/platform/pure-exporter/DEPLOYMENT-SUMMARY.md

Pure FlashArray OpenMetrics Exporter - Deployment Summary

Deployment Date: 2026-06-19
Status: ✅ Successfully Deployed
Namespace: monitoring
Array: utilidor.local.mk-labs.cloud (10.1.71.5)

What Was Deployed

1. Container Deployment

• Image: quay.io/purestorage/pure-fa-om-exporter:v1.0.27 (latest stable)
• Port: 9490 (standard exporter port)
• Resources: 50m CPU / 64Mi memory (requests), 200m CPU / 128Mi memory (limits)
• Health Checks: Liveness and readiness probes on / endpoint
• Security: Running as non-root user (65534)

2. Secret Configuration

• Secret Name: pure-fa-exporter-token
• Contains: tokens.yaml with FlashArray endpoint and API token
• Array ID: utilidor
• Endpoint: utilidor.local.mk-labs.cloud
• API Token: 6157dea0-4c0d-b14b-6e6a-453aa649355d (jarvis readonly user)

3. Service

• Type: ClusterIP (east/west traffic only)
• Port: 9490
• Selector: app.kubernetes.io/name=pure-fa-exporter

4. ServiceMonitor

• Scrape Interval: 60s (storage metrics low frequency)
• Scrape Timeout: 30s
• Path: /metrics?endpoint=utilidor
• Label Selector: release=monitoring (for kube-prometheus-stack)
• Relabeling: Adds 'array' label from endpoint parameter

5. ArgoCD Application

• Wave: 3 (monitoring infrastructure)
• Sync Policy: Automated with prune and self-heal
• Source: cluster/platform/pure-exporter/
• Health: ✅ Healthy
• Sync Status: ✅ Synced

Verification Results

✅ Pod Status

NAME                               READY   STATUS    RESTARTS   AGE
pure-fa-exporter-7d45dcfb7-9rxj2   1/1     Running   0          21s

✅ Service Endpoint

NAME               TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)    AGE
pure-fa-exporter   ClusterIP   10.111.204.251   <none>        9490/TCP   29s

✅ Metrics Available

Successfully scraped FlashArray metrics including:

• purefa_info - System information (Purity//FA 6.5.13)
• purefa_alerts_open - Alert events
• purefa_array_performance_* - Performance metrics (IOPS, bandwidth, latency)
• purefa_array_space_* - Capacity metrics
• purefa_volume_* - Volume metrics
• purefa_host_* - Host metrics
• And many more...

Sample output:

purefa_info{array_name="utilidor",os="Purity//FA",subscription_type="FlashArray",system_id="b78a62d3-57fc-45ab-b73f-6203fb30f28d",version="6.5.13"} 1

Files Created

All manifests committed to homelab repo:

1. application.yaml - ArgoCD Application definition (wave 3)
2. deployment.yaml - Exporter deployment with health checks and resource limits
3. secret.yaml - FlashArray API credentials (tokens.yaml format)
4. service.yaml - ClusterIP service exposing /metrics
5. servicemonitor.yaml - Prometheus Operator integration

Git Commit: cb5ffc1 - "Add Pure FlashArray OpenMetrics exporter to monitoring namespace"

Next Steps for JARVIS

1. Verify Prometheus Scraping

   • Check Prometheus UI → Status → Targets
   • Look for serviceMonitor/monitoring/pure-fa-exporter/0 target
   • Should show state: UP

2. Test Metrics in Prometheus

   • Navigate to Prometheus UI → Graph
   • Query: purefa_info - Should return array information
   • Query: purefa_array_space_bytes{space="capacity"} - Total capacity
   • Query: purefa_array_performance_throughput_iops - IOPS metrics

3. Grafana Dashboard (Optional)

   • The Pure Storage community provides Grafana dashboards
   • Reference: https://github.com/PureStorage-OpenConnect/pure-fa-openmetrics-exporter/tree/master/extra/grafana
   • Can import dashboard JSON for fleet-wide overview

Configuration Notes

Multi-Target Exporter Pattern

The exporter follows Prometheus multi-target pattern:

• Single exporter instance can scrape multiple arrays
• Array selection via ?endpoint=<array_id> query parameter
• Array credentials in tokens.yaml (array_id → endpoint + api_token mapping)

Scrape Frequency

• Current: 60s interval
• Rationale: Storage metrics are relatively static, high frequency unnecessary
• Adjustable: Modify servicemonitor.yaml if faster updates needed

Security Context

• Runs as non-root user (UID 65534)
• Read-only token mount
• No elevated privileges required

Troubleshooting

If metrics are not appearing in Prometheus:

1. Check ServiceMonitor Selection

   kubectl get servicemonitor -n monitoring pure-fa-exporter -o yaml
   

   Ensure release: monitoring label exists for prometheus-operator selector

2. Verify Exporter Logs

   kubectl logs -n monitoring -l app.kubernetes.io/name=pure-fa-exporter
   

3. Test Endpoint Directly

   kubectl run curl-test --image=curlimages/curl:latest --rm -i --restart=Never -n monitoring -- \
     curl -s http://pure-fa-exporter:9490/metrics?endpoint=utilidor | head
   

4. Check Prometheus Operator Logs

   kubectl logs -n monitoring -l app.kubernetes.io/name=prometheus-operator
   

References

• Official Repo: https://github.com/PureStorage-OpenConnect/pure-fa-openmetrics-exporter
• Metrics Specification: See repo for full purefa_* metric namespace documentation
• Deployment Pattern: Based on portworx-csi GitOps structure
• ServiceMonitor Pattern: Based on kube-prometheus-stack conventions

---

Deployment completed successfully. Prometheus should begin scraping FlashArray metrics within 60 seconds.