Prometheus Metrics Exporter for Moesif

A Prometheus exporter that pulls API analytics data from Moesif via its Management API and exposes them as Prometheus metrics.

If you already use Moesif to capture API traffic, this exporter lets you bring that data into Prometheus and Grafana without any additional instrumentation.

How It Works

Prometheus  --scrape-->  This Exporter  --query-->  Moesif Management API
  (pull)                  /metrics        (REST)     (your existing data)

Each time Prometheus scrapes the /metrics endpoint, the exporter queries Moesif for data within a rolling time window (default: last 60 seconds) and returns the results in Prometheus exposition format.

Prerequisites

• Node.js >= 18.0.0
• A Moesif account with API traffic data
• A Moesif Management API Key

Getting Your Moesif Management API Key

1. Log in to your Moesif Dashboard
2. Go to Settings (bottom-left gear icon)
3. Select Management API Keys
4. Create a new key with the following Analytics Read permissions:

The key needs Read access on these Analytics resources:

• companies
• customer_actions
• events
• metrics
• reports
• users

5. Copy your key

Quick Start

# Clone the repo
git clone <repo-url>
cd prometheus-metrics-moesif

# Install dependencies
npm install

# Configure
cp .env.example .env
# Edit .env and add your Moesif Management API Key

# Start the exporter
npm start

The exporter starts on port 9277 by default. Verify it's running:

curl http://localhost:9277/metrics

Configuration

All configuration is done via environment variables or a .env file.

Variable	Required	Default	Description
MOESIF_MANAGEMENT_API_KEY	Yes	—	Your Moesif Management API key
MOESIF_API_BASE_URL	No	https://api.moesif.com	Moesif API base URL
PORT	No	9277	Port the exporter listens on
MOESIF_QUERY_WINDOW_SECONDS	No	60	How far back (in seconds) to query Moesif on each scrape
MOESIF_QUERY_DELAY_SECONDS	No	0	Shift the query window back by this many seconds (see Data Consistency)

Prometheus Configuration

Add a scrape job to your prometheus.yml:

scrape_configs:
  - job_name: 'moesif'
    scrape_interval: 60s
    static_configs:
      - targets: ['localhost:9277']

Important: The scrape_interval should match MOESIF_QUERY_WINDOW_SECONDS so that each scrape covers exactly the time period since the last scrape, with no gaps or overlaps.

For performance reasons, we do not recommend smaller number than 60s.

Metrics Reference

All metrics are exposed as Prometheus gauges (except moesif_scrape_errors_total which is a counter). They represent values within the configured query window.

Global Metrics

Metric	Type	Description
moesif_api_calls_total	Gauge	Total API calls in the query window
moesif_api_calls_by_status	Gauge	API calls by HTTP status code
moesif_api_latency_p50_ms	Gauge	50th percentile latency (ms)
moesif_api_latency_p90_ms	Gauge	90th percentile latency (ms)
moesif_api_latency_p99_ms	Gauge	99th percentile latency (ms)
moesif_active_users	Gauge	Unique users in the query window
moesif_active_companies	Gauge	Unique companies in the query window

Per-Route Metrics

Metric	Labels	Type	Description
moesif_api_calls_by_route	route, status_code	Gauge	API calls by route and status code
moesif_api_latency_by_route_ms	route, percentile	Gauge	Latency percentiles (p50/p90/p99) by route

Operational Metrics

Metric	Type	Description
moesif_scrape_errors_total	Counter	Number of failed Moesif API queries
moesif_scrape_duration_ms	Gauge	Time taken to query Moesif (ms)

Data Consistency and Delay

Moesif uses a distributed architecture with eventual consistency. This means the most recent few seconds of data may not be fully available when queried. Events go through ingestion, processing, and indexing, so there is a small delay before they appear in query results.

By default, the exporter queries from [now - window, now]. If you find that recent data is incomplete or inconsistent, you can shift the entire query window back by setting MOESIF_QUERY_DELAY_SECONDS:

Without delay (default):   |-------- 60s --------|
                                                  now

With 10s delay:            |-------- 60s --------|
                                            now-10s    now
                                              ^         ^
                                          query end   (skipped)

For example, with MOESIF_QUERY_DELAY_SECONDS=10 and MOESIF_QUERY_WINDOW_SECONDS=60, each scrape queries the window [now-70s, now-10s], giving Moesif an extra 10 seconds for data to arrive and be indexed.

A delay of 10-30 seconds is recommended for most deployments. The tradeoff is slightly less real-time data in exchange for more accurate and complete metrics.

Why Gauges Instead of Counters?

Prometheus counters are monotonically increasing and work with rate(). However, this exporter queries Moesif for a rolling time window on each scrape rather than maintaining cumulative state. The returned values can go up or down between scrapes (e.g., a busy minute vs. a quiet minute), so gauges are the correct type.

In Grafana, use the gauge values directly instead of wrapping them in rate().

Example Grafana Queries

# Total API calls per scrape window
moesif_api_calls_total

# 5xx errors by status code
moesif_api_calls_by_status{status_code=~"5.."}

# p90 latency
moesif_api_latency_p90_ms

# Top routes by traffic
topk(10, moesif_api_calls_by_route)

# p99 latency for a specific route
moesif_api_latency_by_route_ms{route="/api/v1/users", percentile="p99"}

# Active users over time
moesif_active_users

Endpoints

Path	Description
/metrics	Prometheus metrics endpoint
/health	Health check (returns {"status": "ok"})

Development

# Run with auto-reload on file changes (Node.js 18+)
npm run dev