Kong Event Gateway metrics

Uses: Event Gateway

Version:

`http.client.request.duration`

Type	Unit
Histogram	`s`

Duration of outbound HTTP client requests.

Labels:

kong.http.client.name: Logical name of the outbound HTTP client, set by the caller to tell multiple clients apart.
http.request.method: HTTP request method. Any method outside the known set maps to _OTHER. (one of CONNECT, DELETE, GET, HEAD, OPTIONS, PATCH, POST, PUT, TRACE)
server.address: Host name of the server the request was sent to.
server.port: Port of the server the request was sent to.
url.scheme: URL scheme of the request, e.g. https. (one of http, https)
network.protocol.version: Negotiated HTTP version of the response. Absent when the request did not succeed. (one of 1.0, 1.1, 2, 3)
http.response.status_code: The status code of an http response. Absent if a request did not succeed.
error.type: The error type encountered on executing http request. Absent if a request was successful. (one of timeout, connect, unknown)

`kong.capability_manager.batches_sent`

Type	Unit
Counter	`{batch}`

Total number of signal batches sent to the client.

Labels:

kong.control_plane: The Control Plane URL a source is polling.

`kong.capability_manager.channel_send_retries`

Type	Unit
Counter	`{retry}`

Total number of retries sending batches to the client channel (channel full).

Labels:

kong.control_plane: The Control Plane URL a source is polling.

`kong.capability_manager.client_processing_duration`

Type	Unit
Histogram	`s`

Time between sending batch to client and receiving acknowledgment.

Labels:

kong.control_plane: The Control Plane URL a source is polling.

`kong.capability_manager.consecutive_failures`

Type	Unit
Gauge	`{failure}`

Current number of consecutive poll failures.

Labels:

kong.control_plane: The Control Plane URL a source is polling.

`kong.capability_manager.signals_received`

Type	Unit
Counter	`{signal}`

Total number of signals received from the control plane (before filtering).

Labels:

kong.control_plane: The Control Plane URL a source is polling.
kong.capability.type: The capability type (e.g. config_sync).

`kong.capability_manager.signals_sent`

Type	Unit
Counter	`{signal}`

Total number of signals sent to the client (after filtering).

Labels:

kong.control_plane: The Control Plane URL a source is polling.

`kong.core.task_registry.tracked_tasks`

Type	Unit
Gauge	`{task}`

Number of active tasks currently tracked by the task tracker.

Labels:

None.

`kong.dataplane.analytics.events`

Type	Unit
Counter	`{event}`

Analytics events handed to the pipeline.

Labels:

None.

`kong.dataplane.analytics.messages.sent`

Type	Unit
Counter	`{message}`

Analytics batches written to the backend.

Labels:

None.

`kong.dataplane.analytics.sent`

Type	Unit
Counter	`By`

Compressed analytics payload bytes written to the backend.

Labels:

None.

`kong.dataplane.analytics.serializer.errors`

Type	Unit
Counter	`{event}`

Errors on the analytics serializer queue.

Labels:

None.

`kong.dataplane.analytics.websocket.errors`

Type	Unit
Counter	`{error}`

Errors on an established analytics websocket.

Labels:

None.

`kong.dataplane.lifecycle.component.ready`

Type	Unit
Gauge	`{state}`

Is a specific component ready; the service being ready implies all components are ready.

Labels:

kong.dataplane.component: The component name

`kong.dataplane.lifecycle.service.healthy`

Type	Unit
Gauge	`{state}`

Is the service healthy.

Labels:

None.

`kong.dataplane.lifecycle.service.ready`

Type	Unit
Gauge	`{state}`

Is the service ready.

Labels:

None.

`kong.keg.config.errors`

Type	Unit
Counter	`{error}`

Count of errors when loading the config received from the control plane.

Labels:

None.

`kong.keg.config.loaded`

Type	Unit
Gauge	`1`

Version number of the configuration loaded from the control plane. Updated each time the control plane pushes a new config.

Labels:

None.

`kong.keg.kafka.acl.attempts`

Type	Unit
Counter	`{attempt}`

Counts the results of every ACL attempt.

Labels:

kong.keg.acl.resource_type: The type of Kafka resource being accessed (one of transactional_id, group, topic, cluster)
kong.keg.result: The result of the ACL check (allowed or denied) (one of allowed, denied)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.auth.virtual_cluster.processing.duration`

Type	Unit
Histogram	`s`

The time spent validating client credentials against virtual cluster auth rules. Passthrough rules are always considered successful even if the backend cluster rejects the authentication.

Labels:

kong.keg.auth.mechanism: The authentication mechanism used by the client
kong.keg.result: The result of the operation (success or failure) (one of success, fail)
kong.keg.auth.mediation: How the proxy mediates authentication between the client and the backend cluster. Empty when no rule matched the requested mechanism. (one of passthrough, validate_forward, terminate, ``)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.backend.connection.errors`

Type	Unit
Counter	`{connection}`

The number of connection errors to the backend cluster.

Labels:

kong.keg.connection.error.origin: The origin of the connection error (one of io, peer, local)

`kong.keg.kafka.backend.roundtrip.duration`

Type	Unit
Histogram	`s`

The time spent communicating with backend cluster (send request and receive response).

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.connection.errors`

Type	Unit
Counter	`{connection}`

The number of proxied connections that resulted in an error.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.listener.id: The Konnect listener identifier
kong.konnect.listener.name: The Konnect listener name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.connections`

Type	Unit
Gauge	`{connection}`

The number of active proxied connections.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.listener.id: The Konnect listener identifier
kong.konnect.listener.name: The Konnect listener name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.decrypt.attempts`

Deprecated. Use kong.keg.kafka.policy.invocations instead.

Type	Unit
Counter	`{attempt}`

The number of attempts to decrypt records. This includes both successful and failed calls

Labels:

kong.keg.result: The result of the operation (success or failure) (one of success, fail)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.encrypt.attempts`

Deprecated. Use kong.keg.kafka.policy.invocations instead.

Type	Unit
Counter	`{attempt}`

The number of attempts to encrypt records. This includes both successful and failed calls

Labels:

kong.keg.result: The result of the operation (success or failure) (one of success, fail)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.metadata.update.duration`

Type	Unit
Histogram	`s`

The time to update the metadata from the backend broker.

Labels:

kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.namespace.topic.conflict`

Type	Unit
Gauge	`1`

Indicates whether the namespace topic mapping encountered conflicts (1) or not (0).

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name

`kong.keg.kafka.policy.condition.failures`

Type	Unit
Counter	`{failure}`

The number of times the policy condition failed to execute due to an error.

Labels:

kong.konnect.policy.id: The Konnect policy identifier
kong.konnect.policy.name: The Konnect policy name
kong.keg.component: The component name

`kong.keg.kafka.policy.invocation.duration`

Type	Unit
Histogram	`s`

The time to process a policy.

Labels:

kong.keg.failure_mode: The configured policy failure mode (one of error, reject, passthrough, mark, skip)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.policy.invocations`

Type	Unit
Counter	`{invocation}`

The number of policy invocation for policies. This includes both successful and failed invocations.

Labels:

kong.keg.failure_mode: The configured policy failure mode (one of error, reject, passthrough, mark, skip)
kong.keg.result: The result of the operation (success or failure) (one of success, fail)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.proxy.duration`

Type	Unit
Histogram	`s`

The end-to-end time for the entire proxy operation.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.request.processing.duration`

Type	Unit
Histogram	`s`

The time spent processing the received request before forwarding it to the backend cluster.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.request.received`

Type	Unit
Counter	`{request}`

The number of requests coming from the client.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.request.sent`

Type	Unit
Counter	`{request}`

The number of requests sent to the backend broker.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.processing.duration`

Type	Unit
Histogram	`s`

The time spent processing the received response before forwarding it to the client.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.received`

Type	Unit
Counter	`{response}`

The number of responses received from the backend broker.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.received.errors`

Deprecated. Use kong.keg.kafka.response.received_errors instead: the old name collided with kong.keg.kafka.response.received as a metric namespace.

Type	Unit
Counter	`{response}`

The number of response messages received from the backend that contain at least one error. The error_code label represents the lowest error in the response.

Labels:

kong.keg.kafka.error_code: The lowest error code in the response
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.received_errors`

Type	Unit
Counter	`{response}`

The number of response messages received from the backend that contain at least one error. The error_code label represents the lowest error in the response.

Labels:

kong.keg.kafka.error_code: The lowest error code in the response
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.sent`

Type	Unit
Counter	`{response}`

The number of responses sent by the proxy to the client.

Labels:

kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.sent.errors`

Deprecated. Use kong.keg.kafka.response.sent_errors instead: the old name collided with kong.keg.kafka.response.sent as a metric namespace.

Type	Unit
Counter	`{response}`

The number of response messages sent back to the client that contain at least one error.

Labels:

kong.keg.kafka.error_code: The lowest error code in the response
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.response.sent_errors`

Type	Unit
Counter	`{response}`

The number of response messages sent back to the client that contain at least one error.

Labels:

kong.keg.kafka.error_code: The lowest error code in the response
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.schema.validation.attempts`

Deprecated. Use kong.keg.kafka.policy.invocations instead.

Type	Unit
Counter	`{attempt}`

The number of attempts to validate schema. This includes both successful and failed calls

Labels:

kong.keg.record.part: The part of the record (key or value) (one of key, value)
kong.keg.result: The result of the operation (success or failure) (one of success, fail)
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.keg.kafka.topic_alias.conflict`

Type	Unit
Gauge	`1`

Emitted (value=1) when a topic alias shadows a physical topic name.

Labels:

kong.keg.kafka.topic_alias.name: The alias name that caused the conflict
kong.keg.kafka.topic_alias.shadowed_topic: The physical topic name that is shadowed by the alias
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name

`kong.keg.konnect.analytics.bytes.sent`

Deprecated. Use kong.dataplane.analytics.sent instead.

Type	Unit
Counter	`By`

Total number of analytics bytes sent in binary websocket messages to the analytics endpoint

Labels:

None.

`kong.keg.konnect.analytics.messages.sent`

Deprecated. Use kong.dataplane.analytics.messages.sent instead.

Type	Unit
Counter	`{message}`

Total number of analytics messages sent to the analytics endpoint.

Labels:

None.

`kong.keg.konnect.analytics.queue.dropped`

Deprecated. Use otel.sdk.processor.log.processed with attribute error.type="queue_full" instead.

Type	Unit
Counter	`{event}`

Number of events dropped from the queue because the max queue size was reached.

Labels:

None.

`kong.keg.konnect.analytics.queue.events`

Deprecated. Use kong.dataplane.analytics.events instead.

Type	Unit
Counter	`{event}`

Total number of events added to the queue.

Labels:

None.

`kong.keg.konnect.analytics.sent`

Deprecated. Use kong.dataplane.analytics.sent instead.

Type	Unit
Counter	`By`

Total bytes sent in binary websocket messages to the analytics endpoint.

Labels:

None.

`kong.keg.konnect.analytics.websocket.errors`

Deprecated. Use kong.dataplane.analytics.websocket.errors instead.

Type	Unit
Counter	`{error}`

Number of times an error occurred on the analytics websocket connection while sending or receiving messages.

Labels:

None.

`kong.keg.konnect.request.duration`

Deprecated. Use http.client.request.duration with the kong.http.client.name label instead.

Type	Unit
Histogram	`s`

The time sending and receiving the response to a request to the Konnect control plane.

Labels:

kong.keg.konnect.api: The konnect api operation being performed (one of fetch_config, update_dp_state)
error.type: The error type encountered on executing http request. Absent if a request was successful. (one of timeout, connect, unknown)
http.response.status_code: The status code of an http response. Absent if a request did not succeed.

`kong.keg.lifecycle.component.ready`

Deprecated. Use kong.dataplane.lifecycle.component.ready instead.

Type	Unit
Gauge	`{state}`

Is a specific component ready; the service being ready implies all components are ready.

Labels:

kong.keg.component: The component name

`kong.keg.lifecycle.service.healthy`

Deprecated. Use kong.dataplane.lifecycle.service.healthy instead.

Type	Unit
Gauge	`{state}`

Is the service healthy.

Labels:

None.

`kong.keg.lifecycle.service.ready`

Deprecated. Use kong.dataplane.lifecycle.service.ready instead.

Type	Unit
Gauge	`{state}`

Is the service ready.

Labels:

None.

`kong.keg.listener.connections.limit`

Type	Unit
Gauge	`{connection}`

The number of allowed connections to the listener.

Labels:

None.

`kong.keg.request_rule.condition.failures`

Type	Unit
Counter	`{failure}`

The number of times a request rule expression failed to execute due to an error.

Labels:

kong.keg.kafka.request: The type of the Kafka request
kong.konnect.policy.id: The Konnect policy identifier
kong.konnect.policy.name: The Konnect policy name
kong.keg.component: The component name

`kong.keg.request_rule.evaluations`

Type	Unit
Counter	`{evaluation}`

Counts the result of every request rule evaluation. Incremented once per evaluated rule.

Labels:

kong.keg.request_rule.result: The result of the request rule evaluation (pass or violation) (one of pass, violation)
kong.keg.request_rule.action: The configured action of the evaluated rule (one of reject, passthrough)
kong.keg.kafka.request: The type of the Kafka request
kong.konnect.policy.id: The Konnect policy identifier
kong.konnect.policy.name: The Konnect policy name
kong.konnect.virtual_cluster.id: The Konnect virtual cluster identifier
kong.konnect.virtual_cluster.name: The Konnect virtual cluster name
kong.konnect.backend_cluster.id: The Konnect backend cluster identifier
kong.konnect.backend_cluster.name: The Konnect backend cluster name
kong.keg.component: The component name

`kong.tokio.global_queue.depth`

Type	Unit
UpDownCounter	`{task}`

Tasks waiting in a Tokio runtime’s global queue.

Labels:

kong.tokio.runtime: The Tokio runtime the value belongs to: main for the dataplane’s #[tokio::main] one, otlp_export for the OTLP exporters’ one, or another runtime’s own name.

`kong.tokio.task.count`

Type	Unit
UpDownCounter	`{task}`

Tasks alive in a Tokio runtime.

Labels:

kong.tokio.runtime: The Tokio runtime the value belongs to: main for the dataplane’s #[tokio::main] one, otlp_export for the OTLP exporters’ one, or another runtime’s own name.

`kong.tokio.worker.busy.time`

Type	Unit
Counter	`s`

Time a Tokio worker thread has spent processing work.

Labels:

kong.tokio.runtime: The Tokio runtime the value belongs to: main for the dataplane’s #[tokio::main] one, otlp_export for the OTLP exporters’ one, or another runtime’s own name.
kong.tokio.worker: Index of the worker thread within its Tokio runtime, an integer.

`kong.tokio.worker.count`

Type	Unit
UpDownCounter	`{thread}`

Worker threads of a Tokio runtime.

Labels:

kong.tokio.runtime: The Tokio runtime the value belongs to: main for the dataplane’s #[tokio::main] one, otlp_export for the OTLP exporters’ one, or another runtime’s own name.

`otel.sdk.processor.log.errors`

Deprecated. Use otel.sdk.processor.log.processed with attribute error.type="queue_full" instead.

Type	Unit
Counter	`{log_record}`

The number of log records dropped because the log processor’s queue was full.

Labels:

None.

Kong Event Gateway metrics

Help us make these docs great!

Still need help?