Prometheus Metrics — Aralez — Reverse Proxy built on Cloudflare's Pingora

Prometheus metrics are exposed at http://config_address/metrics — by default http://127.0.0.1:3000/metrics.

Example Grafana Dashboard

Grafana Dashboard during stress test

Metric Name	Type	Unit	Description
`aralez_active_sessions`	Gauge	Count	Current number of active user/client sessions.
`aralez_cache_evicted_bytes_total`	Gauge	Bytes	Total memory size of data evicted from the cache.
`aralez_cache_evicted_items_total`	Gauge	Count	Total number of individual items evicted from the cache.
`aralez_cache_items`	Gauge	Count	Current total number of items stored in the cache.
`aralez_cache_size_bytes`	Gauge	Bytes	Current memory consumption of the cache in bytes.
`aralez_logging_errors`	Gauge / Counter	Count	Total count of logging error events recorded.
`aralez_memory_bytes`	Gauge	Bytes	Total memory currently allocated by the process in bytes.
`aralez_open_files`	Gauge	Count	Number of open file descriptors currently held by the application.
`aralez_requests_by_method_total`	Counter	Count	Total HTTP requests processed, partitioned by HTTP method.
`aralez_requests_by_upstream`	Counter	Count	Total HTTP requests routed, partitioned by targeted upstream server.
`aralez_requests_by_version_total`	Counter	Count	Total HTTP requests received, partitioned by HTTP protocol version.
`aralez_requests_total`	Counter	Count	Aggregate count of all HTTP requests handled by Aralez.
`aralez_response_latency_seconds`	Histogram	Seconds	Distribution of HTTP response latency in seconds.
`aralez_responses_total`	Counter	Count	Total HTTP responses returned, partitioned by HTTP status code.

Type: Counter
Description: Tracks the cumulative number of HTTP requests processed by the Aralez application since process start.
Usage: Calculate overall throughput using PromQL: rate(aralez_requests_total[5m]).

Type: Counter
Labels:
- method: The HTTP request method (e.g., GET, POST, PUT, HEAD).
Description: Tracks request traffic broken down by HTTP protocol verb.

Type: Counter
Labels:
- version: The HTTP protocol version used (e.g., HTTP/1.1, HTTP/2.0).
Description: Breakdown of request volume by protocol version.

Type: Counter
Labels:
- upstream: Hostname or IP of the destination upstream server (e.g., localhost, apt.netangels.net).
Description: Tracks traffic distribution across upstream backend dependencies.

Type: Counter
Labels:
- status: HTTP response status code (e.g., 200, 502).
Description: Total responses generated, grouped by HTTP response code.
Usage: Useful for calculating error rates (e.g., ratio of 5xx responses to total responses).

Type: Histogram
Associated Metrics:
- aralez_response_latency_seconds_bucket{le="..."}: Count of requests completed within the specified latency threshold (le = less than or equal to seconds).
- aralez_response_latency_seconds_sum: Cumulative total of all response latencies in seconds.
- aralez_response_latency_seconds_count: Total number of measured requests (matches aralez_requests_total).
Description: Measures request execution duration to calculate percentiles (p50, p95, p99).
Usage: Calculate 95th percentile latency over 5 minutes: histogram_quantile(0.95, sum(rate(aralez_response_latency_seconds_bucket[5m])) by (le))

Type: Gauge
Description: Instantaneous measurement of the number of items currently held in memory cache.

Type: Gauge (Behaves as Counter)
Description: Cumulative number of objects removed from the cache due to memory limits or eviction policies.

Type: Gauge (Behaves as Counter)
Description: Cumulative size (in bytes) of all objects evicted from the cache.

Type: Gauge
Description: Real-time count of currently connected or active client sessions.

Type: Gauge
Description: Current memory allocated by the application process (in bytes).

Type: Gauge
Description: Number of open file handles currently used by the process. Monitor to prevent hitting operating system file descriptor limits (ulimit).