Skip to content

Latest commit

 

History

History
74 lines (55 loc) · 4.44 KB

enrichment.md

File metadata and controls

74 lines (55 loc) · 4.44 KB
Raw

Enrichment performed by Promregator

Promregator may enrich the scraping results for you. It does so by providing additional metrics.

In general, there are two cases where additional metrics may appear:

  • Additional metrics measuring the communication to the targets.
  • Additional metrics measuring Promregator itself.

Additional Metrics Measuring the Communication to the Targets

Promregator monitors the connectivity to the targets. The result of the monitoring also made available to the caller via Prometheus metrics. For this, the following metrics are exposed:

  • promregator_request_latency: a Prometheus histogram, which returns the latency which was necessary to retrieve the metrics from the target. Note that this metric is disabled by default due to high data volume. You may enable it with configuration option promregator.metrics.requestLatency.
  • promregator_request_size: a Prometheus histogram, which returns the size of the scraping document, which was sent from the target to Promregator.
  • promregator_up: a Prometheus Gauge which indicates whether an instance was reachable or not (similar to the gauge provided for Prometheus' own monitoring).
  • promregator_request_failure: a Prometheus Gauge which indicates the number of requests sent to the target, which have failed.
  • promregator_scrape_duration_seconds: a Prometheus Gauge which indicates how long scraping of the current request target took. This metric is dependent on the chosen target and thus provides the labels org_name, space_name, app_name, cf_instance_number and cf_instance_id (as stated below).

Here is an example how such metric samples may look like:

# HELP promregator_up Indicator, whether the target of promregator is available
# TYPE promregator_up gauge
promregator_up 1.0

# HELP promregator_request_latency The latency, which the targets of the promregator produce
# TYPE promregator_request_latency histogram
promregator_request_latency_bucket{le="0.005"} 0.0
promregator_request_latency_bucket{le="0.01"} 0.0
promregator_request_latency_bucket{le="0.025"} 0.0
promregator_request_latency_bucket{le="0.05"} 0.0
promregator_request_latency_bucket{le="0.075"} 0.0
promregator_request_latency_bucket{le="0.1"} 0.0
promregator_request_latency_bucket{le="0.25"} 1.0
promregator_request_latency_bucket{le="0.5"} 1.0
promregator_request_latency_bucket{le="0.75"} 1.0
promregator_request_latency_bucket{le="1.0"} 1.0
promregator_request_latency_bucket{le="2.5"} 1.0
promregator_request_latency_bucket{le="5.0"} 1.0
promregator_request_latency_bucket{le="7.5"} 1.0
promregator_request_latency_bucket{le="10.0"} 1.0
promregator_request_latency_bucket{le="+Inf"} 1.0
promregator_request_latency_count 1.0
promregator_request_latency_sum 0.21851916

Under certain circumstances (i.e. HA Setup) label enrichment (e.g. for label names orgName, spaceName, appName) may collide with Prometheus' Relabeling. In that case, you may make Promregato prefix all its label names for internal metrics by using configuration option promregator.metrics.labelNamePrefix.

Additional Metrics Measuring Promregator Itself

The metrics provided by Promregator also supports you in monitoring Promregator itself. Due to this, additional metrics, such as the Hotspot's (Java Virtual Machine) metrics are also exposed to the caller.

Note that, as these metrics also are originated in Promregator, they are always prefixed with promregator_.

See here some examples, how this may look like:

# HELP promregator_jvm_memory_pool_bytes_max Max bytes of a given JVM memory pool.
# TYPE promregator_jvm_memory_pool_bytes_max gauge
promregator_jvm_memory_pool_bytes_max{pool="Code Cache"} 2.5165824E8
promregator_jvm_memory_pool_bytes_max{pool="Metaspace"} -1.0
promregator_jvm_memory_pool_bytes_max{pool="Compressed Class Space"} 1.073741824E9
promregator_jvm_memory_pool_bytes_max{pool="PS Eden Space"} 1.358430208E9

By default, only the metrics of the the Java Hotspot are exposed like this. Using the configuration option promregator.metrics.internal further internal metrics may be enabled.