io.confluent.kafka.multitenant.metrics

Class ConnectionInformationSensorBuilder

java.lang.Object

io.confluent.kafka.multitenant.metrics.ConnectionInformationSensorBuilder

public class ConnectionInformationSensorBuilder
extends Object

Constructor Summary

Constructors

Constructor and Description
ConnectionInformationSensorBuilder(org.apache.kafka.common.metrics.Metrics metrics, io.confluent.kafka.multitenant.MultiTenantPrincipal principal, org.apache.kafka.common.network.ClientInformation clientInformation, String clientId, int clientIdsPerTenantCap)

Method Summary

Modifier and Type	Method and Description
org.apache.kafka.common.metrics.Sensor	maybeBuildAndRecord() The value of this metric isn't in the count, but the tags, which are used by the UI to get a list of clients per tenant.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ConnectionInformationSensorBuilder

public ConnectionInformationSensorBuilder(org.apache.kafka.common.metrics.Metrics metrics,
                                          io.confluent.kafka.multitenant.MultiTenantPrincipal principal,
                                          org.apache.kafka.common.network.ClientInformation clientInformation,
                                          String clientId,
                                          int clientIdsPerTenantCap)

Method Detail

maybeBuildAndRecord

public org.apache.kafka.common.metrics.Sensor maybeBuildAndRecord()

The value of this metric isn't in the count, but the tags, which are used by the UI to get a list of clients per tenant. To avoid exhausting memory, we only want to record up to clientIdsPerTenantCap uniques. We do it by only having another metric of unique client ids per tenant, which we also record if we don't find an existing sensor for this client ID, and verify quotas on it if we do.

Returns:

the sensor, after record(), if we're under the cap, or null if we're over.