Exposing IBM Storage Scale (GPFS) Metrics to IBM Cloud Monitoring Using Sysdig Agent - IBM/ibm-spectrum-scale-bridge-for-grafana GitHub Wiki

IBM Storage Scale exposes cluster performance data through the internal ZiMon performance monitoring subsystem.

The IBM Storage Scale Bridge for Grafana bridges this gap. It reads raw ZiMon metric data and re-exposes it in Prometheus exposition format on port 9250. The Sysdig Agent (dragent - the agent process managing metric collection and forwarding to IBM Cloud Monitoring) - already deployed - scrapes that endpoint and forwards every GPFS metric to your IBM Cloud Monitoring instance.

This gives you a single-pane-of-glass view for GPFS disk I/O, CPU, NSD, filesystem throughput and cluster health - with alerting and capacity planning - without needing a standalone Grafana or Prometheus server.

Architecture & Data Flow

 ┌─────────────────────────────────────────────────────────────────┐
 │                  IBM Storage Scale Cluster                      │
 │                                                                 │
 │  ┌─────────────────────────────────────┐                        │
 │  │         Management Node             │                        │
 │  │                                     │                        │
 │  │  ┌──────────────────────────────┐   │                        │
 │  │  │  ZiMon / pmcollector         │   │                        │
 │  │  │  REST API  :9980 (HTTPS)     │   │                        │
 │  │  └──────────┬───────────────────┘   │                        │
 │  │             │  queries / JSON        │                        │
 │  │             ▼                        │                        │
 │  │  ┌──────────────────────────────┐   │                        │
 │  │  │  Grafana Bridge              │   │                        │
 │  │  │  Prometheus exporter :9250   │   │                        │
 │  │  │  (HTTPS + Basic Auth + TLS)  │   │                        │
 │  │  └──────────┬───────────────────┘   │                        │
 │  │             │  scrape /metrics       │                        │
 │  │             ▼                        │                        │
 │  │  ┌──────────────────────────────┐   │                        │
 │  │  │  Sysdig dragent              │   │                        │
 │  │  │  (promscrape.yaml.d/)        │   │                        │
 │  │  └──────────────────────────────┘   │                        │
 │  └─────────────────────────────────────┘                        │
 │                                                                 │
 │  ┌──────────────────────┐                                       │
 │  │  Worker Nodes (N)    │                                       │
 │  │  Sysdig dragent      │  ← host metrics only (CPU, mem, net) │
 │  └──────────────────────┘                                       │
 └─────────────────────────────────────────────────────────────────┘
          │  GPFS + host metrics (HTTPS :6443)
          ▼
 ┌─────────────────────────────────┐
 │       IBM Cloud Monitoring      │
 │  Ingestion endpoint  :6443      │
 │  Dashboards · Alerts · Metrics  │
 └─────────────────────────────────┘

Key Components

Grafana Bridge

Runs on the pmcollector node only. Translates ZiMon data into Prometheus format. Also auto-generates a ready-to-use scrape config at GET /prometheus.yml covering all GPFS sensor groups (GPFSFilesystem, GPFSNSD, GPFSNetwork etc)

Sysdig dragent

IBM Cloud Monitoring agent. Installed on every scale node. On the management node it is additionally configured with a custom prometheus scrape directory (/opt/draios/etc/promscrape.yaml.d/) to also forward GPFS metrics.

`scale_bridge.yaml` (scrape config injection)

Instead of manually writing Prometheus scrape configs, the automation fetches the auto-generated prometheus.yml directly from the running bridge, extracts the scrape_configs block, and writes it to /opt/draios/etc/promscrape.yaml.d/scale_bridge.yaml The GPFSPDDisk job is stripped for non-ESS clusters.

Verifying the Setup

Bridge is up and serving metrics:

curl -sk -u svc_osprey_scraper:<password> https://127.0.0.1:9250/metrics | head -30

Sysdig scrape config was injected:

cat /opt/draios/etc/promscrape.yaml.d/scale_bridge.yaml

Agent forwarding to IBM Cloud Monitoring:

systemctl status dragent
journalctl -u dragent --since "5 min ago" | grep -i "prometheus\|scrape\|gpfs"

In IBM Cloud Monitoring: filter by metric namespace gpfs_* or tag cluster:ibm_storage_scale. Metrics appear within 60–90 seconds of agent startup.

Example of dragent.yml

[vpcuser@npr-scale27-strg-mgmt-850a-001 promscrape.yaml.d]$ cat /opt/draios/etc/dragent.yaml
customerid: "47e36686-c492-445a-8a4b-278620e90b89"
collector: "ingest.jp-tok.monitoring.cloud.ibm.com"
collector_port: 6443
tags: "cluster:ibm_storage_scale,nodeclass:managementnodegrp"
sysdig_capture_enabled: false
remotefs: true
prometheus:
  enabled: true
  yaml_dir: /opt/draios/etc/promscrape.yaml.d

first two values are endpoints of cloud monitoring instance

Example IBM Cloud Monitoring Views

Scale metrics in Cloud Monitoring

IBM Cloud Monitoring view showing IBM Storage Scale metrics

Cloud Monitoring instance dashboard

IBM Cloud Monitoring instance dashboard view