BTP Cloud Foundry Application Stats Monitor

The BTP Cloud Foundry Application Stats Monitor provides comprehensive monitoring of Cloud Foundry applications deployed on SAP BTP. It tracks application health, resource usage, and instance status in real-time by communicating directly with the Cloud Foundry V3 API.

Prerequisites

Network connectivity from the collector to BTP API endpoints (e.g., api.cf.eu10.hana.ondemand.com).

BTP Credentials: A user with at least Space Auditor permissions for the target spaces.

Endpoints accessed: /v3/apps: To discover applications and their relationships (Space/Org). /v3/apps/{guid}/processes/web/stats: To retrieve real-time resource usage and instance states. Authentication: OAuth2 Password Grant (handled automatically using the configured User Profile).

API Endpoints used

Endpoint	Purpose
POST /oauth/token	Authentication (via UAA login endpoint)
GET /v3/apps	Discovery of all apps with Org/Space details
GET /v3/apps/{guid}/processes/web/stats	Retrieve instance-level statistics

Key Features

Metrics Collection

The monitor collects the following metrics for each application instance: Running Instances - Count of instances in RUNNING state. Crashed Instances - Count of instances in CRASHED or DOWN state. Memory Usage - Current consumption (MB) and quota utilization (%). Disk Usage - Current consumption (MB) and quota utilization (%). CPU Usage - CPU utilization percentage (0-100%).

Alarm Capabilities

State Monitoring - Detect when instances enter undesired states (CRASHED, DOWN, FAILED, UNKNOWN). NOT_RUNNING includes all the failed states.
Resource Thresholds - Multi-level thresholds (Warning, Critical, etc.) for Memory, Disk, and CPU usage using the standard G2W:80 W2M:90 format - Multi thresholds syntax
Auto-Clear - Automatically resolves alarms when the application returns to a healthy state.

Advanced Discovery

Wildcard Support: Use * in the App Name field to monitor all apps matching optional Subaccount/Space filters. Discover All Mode: An optional toggle to collect metrics for every application in the subaccount, even if not explicitly listed in the configuration table (metrics only, no alarms). Parallel Execution: Uses a thread pool to fetch statistics for multiple applications simultaneously, significantly reducing collection time.

How It Works

Data Collection Process

Authentication: Converts the API URL to a Login URL (api.cf… → login.cf…) to fetch an OAuth2 token, which is then cached. Metadata Caching: Discovered applications (Names, GUIDs, Orgs, Spaces) are cached for 1 hour to minimize API overhead. Process Stats: For each app, the monitor specifically targets the web process statistics. State Evaluation: Alarms are raised per-app for state issues and per-instance for resource usage.

Configuration

Connection Settings

Create a Web Service Connector: URL: Your BTP API endpoint (e.g., https://api.cf.eu10.hana.ondemand.com). Auth Type: Ensure the connector is associated with a User Profile containing valid BTP credentials.

Monitor Configuration

Method 1: Load BTP Apps (Recommended)

Open the monitor configuration. Click the “Load BTP Apps” button. The table will populate with the App Name, Subaccount (Org), and Space. Activate the desired rows and configure specific thresholds.

Method 2: Wildcard Mode

App Name: Set to *. Subaccount / Space: Set to * to monitor everything, or enter a specific name to filter the wildcard discovery.

Settings Reference

Field	Description	Default
Active	Enable/disable this configuration row	`true`
App Name	Specific app name or	`*`
Subaccount	Filter by BTP Organization name	`*`
Space	Filter by BTP Space name	`*`
Discover All	Collect metrics for all apps in the subaccount	`true`
Send Metrics	Enable/disable performance data storage	`true`
State Monitor	Choose which state triggers an alarm (e.g., `NOT_RUNNING`)	`NOT_RUNNING`
Memory Quota	Threshold for memory usage (e.g., `G2W:80 W2M:90`)	`80/90`
CPU Usage	Threshold for CPU usage (e.g., `G2W:70 W2M:85`)	`70/85`

Collected Metrics

Metrics are stored with the base name: promonitor.btp.cloud_foundry.applications.*

Metric Key	Unit	Description	Tags
running_instances	count	Healthy instances	app_name, org_name, space_name
crashed_instances	count	Failed/Down instances	app_name, org_name, space_name
memory_used_mb	MB	Memory consumed	app_name, org_name, space_name, instance#
disk_used_mb	MB	Disk consumed	app_name, org_name, space_name, instance#
cpu_usage_percent	%	CPU utilization	app_name, org_name, space_name, instance#

Troubleshooting

“No cache → skipping monitoring”: The monitor failed to load the application list. Check the connector URL and credentials.

HTTP 401/403: Verify the user has the Space Auditor role in BTP.

Metrics show 0 but app is running: Check if the app is using a process type other than web. The monitor currently focuses on the web process stats.

Stale Data: The app list is cached for 60 minutes. If you added a new app, it might take an hour to appear unless you trigger “Load BTP Apps” or the cache expires.

Products documentation

Table of Contents