旧バージョンのドキュメントを参照しています。最新のドキュメントはこちらをご参照ください。

Prometheus Metrics

Kong Ingress Controller, as well as Kong Gateway, both expose Prometheus metrics, under certain conditions:

Kong Ingress Controller version 2.0 and later, exposes Prometheus metrics for configuration updates.
Kong Gateway can expose Prometheus metrics for the requests that are served, if the Prometheus plugin is enabled. For more information about enabling plugin, see Using KongPlugin resource guide. You can also learn about specific integration details in the Integration with Prometheus and Grafana guide.

This is a reference for Integration with Prometheus and Grafana guide.

Prometheus metrics for configuration updates

ingress_controller_configuration_push_count

ingress_controller_configuration_push_count (type: counter) provides the number of successful or failed configuration pushes to Kong Gateway.

This metric provides these labels:

protocol describes the configuration protocol in use, which can be db-less or deck.
dataplane describes the data plane that was the target of configuration push.
success logs the status of configuration updates. If success is false, an unrecoverable error occurred. If success is true, the push succeeded with no errors.
failure_reason is populated if success="false". It describes the reason for the failure:
- conflict: A configuration conflict that must be manually fixed.
- network: A network related issues, such as Kong Gateway is offline.
- other: Other issues, such as Kong Gateway reporting a non-conflict error.

ingress_controller_translation_count

ingress_controller_translation_count (type: counter) provides the number of translations from the Kubernetes state to the Kong Gateway state.

This metric provides the success label. success logs the status of configuration updates. If success is false, an unrecoverable error occurred. If success is true, the translation succeeded with no errors.

ingress_controller_configuration_push_duration_milliseconds

ingress_controller_configuration_push_duration_milliseconds (type: histogram) is the amount of time, in milliseconds, that it takes to push the configuration to Kong Gateway.

This metric provides these labels:

protocol describes the configuration protocol in use, which can be db-less or deck.
dataplane describes the data plane that was the target of configuration push.
success logs the status of configuration updates. If success is false, an unrecoverable error occurred. If success is true, the push succeeded with no errors.

ingress_controller_configuration_push_broken_resource_count

ingress_controller_configuration_push_broken_resource_count (type: counter) provides the number of resources not accepted by Kong Gateway when attempting to push configuration.

This metric provides these labels:

dataplane describes the data plane that was the target of the configuration push

ingress_controller_configuration_push_last_successful

ingress_controller_configuration_push_last_successful (type: gauge) provides the time of the last successful configuration push.

This metric provides these labels:

dataplane describes the data plane that was the target of the configuration push

ingress_controller_translation_broken_resource_count

ingress_controller_translation_broken_resource_count (type: gauge) provides the number of resources that the controller cannot successfully translate to Kong Gateway configuration.

ingress_controller_fallback_translation_count

ingress_controller_fallback_translation_count (type: counter) provides the count of translations from Kubernetes state to Kong state in fallback mode.

This metric provides the success label:

success logs the status of the translation. If success is false, an unrecoverable error occurs. If success is true, the translation succeeded without errors.

ingress_controller_fallback_translation_broken_resource_count

ingress_controller_fallback_translation_broken_resource_count (type: gauge) provides the number of resources that the controller cannot successfully translate to Kong configuration in fallback mode.

ingress_controller_fallback_configuration_push_count

ingress_controller_fallback_configuration_push_count (type: counter) provides the count of successful/failed fallback configuration pushes to Kong.

This metric provides these labels:

dataplane describes the data plane that was the target of the configuration push.
protocol describes the configuration protocol in use, which can be db-less or deck.
success logs the status of configuration updates. If success is false, an unrecoverable error occurs. If success is true, the push succeeded without errors.
failure_reason is populated if success="false". It describes the reason for the failure:
- conflict: A configuration conflict that must be manually fixed.
- network: A network related issue, such as Kong Gateway being offline.
- other: Other issues, such as Kong Gateway reporting a non-conflict error.

ingress_controller_fallback_configuration_push_last

ingress_controller_fallback_configuration_push_last (type: gauge) provides the time of the last successful fallback configuration push.

This metric provides these labels:

dataplane describes the data plane that was the target of the configuration push.

ingress_controller_fallback_configuration_push_duration_milliseconds

ingress_controller_fallback_configuration_push_duration_milliseconds (type: histogram) provides the amount of time, in milliseconds, that it takes to push the fallback configuration to Kong.

This metric provides these labels:

dataplane describes the data plane that was the target of the configuration push.
protocol describes the configuration protocol in use, which can be db-less or deck.
success logs the status of configuration updates. If success is false, an unrecoverable error occurs. If success is true, the push succeeded without errors.

ingress_controller_fallback_configuration_push_broken_resource_count

ingress_controller_fallback_configuration_push_broken_resource_count (type: gauge) provides the number of resources not accepted by Kong when attempting to push the fallback configuration.

This metric provides these labels:

dataplane describes the data plane that was the target of the configuration push.

ingress_controller_fallback_cache_generation_duration_milliseconds

ingress_controller_fallback_cache_generation_duration_milliseconds (type: histogram) provides the amount of time, in milliseconds, that it takes to generate a fallback cache.

This metric provides the success label:

success logs the status of cache generation. If success is false, an unrecoverable error occurs. If success is true, the cache generation succeeded without errors.

ingress_controller_processed_config_snapshot_cache_hit

ingress_controller_processed_config_snapshot_cache_hit (type: counter) provides the count of times the controller hit the processed configuration snapshot cache and skipped generating a new one.

ingress_controller_processed_config_snapshot_cache_miss

ingress_controller_processed_config_snapshot_cache_miss (type: counter) provides the count of times the controller missed the processed configuration snapshot cache and had to generate a new one.

Low-level performance metrics

In addition, Kong Ingress Controller exposes more low-level performance metrics, but these may change from version to version because they are provided by underlying frameworks of Kong Ingress Controller.

A non-exhaustive list of these low-level metrics is described in the following: