Global Profiling Engine

Configure Prometheus metrics for the Global Profiling Engine

This page applies to:

  • Global Profiling Engine 1.0 and newer

The Global Profiling Engine supports Prometheus metrics that reveal characteristics of your stick table data and the health of the GPE service. You could import this data into dashboarding software, such as Grafana, to see trends over time.

Enable metrics Jump to heading

To enable Prometheus metrics:

  1. On the Global Profiling Engine server, edit the file /etc/hapee-extras/hapee-gpe.json.

  2. Set the prometheus_exporter field to one of the following values:

    Value Description
    0 Disable metrics. (default)
    1 Enable only general metrics.
    2 Enable only health metrics.
    3 Enable both general and health metrics.

    In this example, we enable both general and health metrics:

    hapee-gpe.json
    json
    {
    "worker_thread_count": 4,
    "inter_worker_queue_size": 1024,
    "collector_queue_size": 64,
    "httpd_port": 9888,
    "datadir": "/var/cache/hapee-extras/hct_datadir",
    "default_stick_table_handling": 1,
    "prometheus_exporter": 3,
    ...
    hapee-gpe.json
    json
    {
    "worker_thread_count": 4,
    "inter_worker_queue_size": 1024,
    "collector_queue_size": 64,
    "httpd_port": 9888,
    "datadir": "/var/cache/hapee-extras/hct_datadir",
    "default_stick_table_handling": 1,
    "prometheus_exporter": 3,
    ...

    The metrics are available at the port set by httpd_port at /metrics for general metrics and /health for health metrics.

Metrics reference Jump to heading

This section describes the available metrics.

General metrics Jump to heading

General metrics record the count, sum, average, minimum, maximum, and percentiles of your stick table data.

Metric Type Description
gpe_metrics_cnt gauge Count.
gpe_metrics_sum gauge Sum.
gpe_metrics_avg gauge Average.
gpe_metrics_per_sec_avg gauge Per second average.
gpe_metrics_burst_avg gauge Burst average.
gpe_metrics_min gauge Minimum.
gpe_metrics_max gauge Maximum.
gpe_metrics_50p gauge Histogram 50 percentile.
gpe_metrics_75p gauge Histogram 75 percentile.
gpe_metrics_90p gauge Histogram 90 percentile.
gpe_metrics_95p gauge Histogram 95 percentile.
gpe_metrics_99p gauge Histogram 99 percentile.
gpe_metrics_99_9p gauge Histogram 99.9 percentile.

Health metrics Jump to heading

Health metrics reveal the health of your Global Profiling Engine service.

Metric Type Description
gpe_coll_q_usage gauge Usage of the collector queues (%).
gpe_dropped_since_start counter Number of messages dropped since start.
gpe_open_fds gauge Open file descriptors.
gpe_raw_q_usage gauge Usage of the RAW worker queues (%).
gpe_resident_memory_btyes gauge Resident memory size in bytes.
gpe_stat_q_usage gauge Usage of the STAT worker queues (%).
gpe_stktagg_connections gauge Number of connections (stktagg).
gpe_stktagg_tables counter Number of stkt tables (stktagg).
gpe_stktagg_updates_since_start counter Number of update messages received (stktagg) since start.
gpe_total_cpu_time gauge Total user and system CPU time spent in seconds.
gpe_uptime gauge Uptime in seconds.
gpe_virtual_memory_bytes gauge Virtual memory size in bytes.
gpe_writes_since_start counter Number of messages written to worker queues since start.

Do you have any suggestions on how we can improve the content of this page?