Apache Kafka

Apache Kafka

Automatic and intelligent observability with trace and metric insights.

Extension

Free trial Documentation

Product information
Release notes

Overview

With Dynatrace, you can get observability for Kafka without touching any code, thanks to automatic monitoring. Seamless end-to-end traces for connected producer and consumer clients allow you to diagnose anomalies and pinpoint the root cause of the broken code before your customers are affected. Comprehensive metrics give you insight into your Kafka servers with health and performance metrics for brokers, topics, producers, and consumers. Events point you to critical anomalies, reducing the mean repair time.

Use cases

Capture every message across tiers without blind spots.
Improve the performance of your producer and consumer services end-to-end.
Troubleshoot asynchronous service problems across your stack proactively.
Prevent message processing anomalies to reduce the mean time to repair.
Monitor the health and performance of all your brokers and topics.
Understand your consumer lag.

Get started

Messaging clients (applications)

To get trace insight:

Install OneAgent on the virtual machine or server of your messaging clients (applications).
Set up Dynatrace on Kubernetes or OpenShift for your messaging client (application) workloads.
Activate the following OneAgent features:
- Java Kafka
- Java Kafka Streams
- Spring for Apache Kafka
- Node.js KafkaJs
- .NET Messaging Apache Kafka

To get log insight:

Activate log monitoring.

Messaging servers (brokers)

Prerequisites:

Kafka broker, producer and consumer running on a supported Linux or Windows operating system.
OneAgent version 1.270+
Dynatrace version 1.270+

To get metric insight:

Install OneAgent on the virtual machine or server of your Kafka broker process.
Ensure your Kafka broker, producer and consumer processes are monitored.
Activate the OneAgent feature Java Metric Extensions 2.0 (JMX).
Select in Dynatrace Add to environment to configure the extension.
Open the Apache Kafka Overview dashboard.

Extension

With the Kafka extension, you can get additional insight into your Kafka server with metrics for brokers, topics, producers, consumers, and more. The extension also provides alerts for the most critical metrics. It creates a custom topology and entities for brokers, topics, producers, and consumers. It provides a dashboard for easier access and configuration of the extension and its entities.

The extension will gather different metrics depending on if it's monitoring a Kafka broker, producer or consumer. Make sure to activate the extension on all of them to get all metrics.

Requirements

OneAgent version 1.270+
Dynatrace version 1.270+
OneAgent feature Java Metric Extensions 2.0 (JMX) activated

Select Add to environment to get started.

Subscribe to new releases

Copy to clipboard

Extension content

Content typeNumber of items included

alerts

4

metric metadata

74

list screen layout

4

screen message cards

4

screen layout

4

screen injections

6

screen chart groups

8

screen actions

8

screen metric tables

4

screen entities lists

11

dashboards

1

generic relationship

8

generic type

4

screen properties

3

Feature sets

Below is a complete list of the feature sets provided in this version. To ensure a good fit for your needs, individual feature sets can be activated and deactivated by your administrator during configuration.

Feature setsNumber of metrics included

Metric name	Metric key	Description	Unit
Kafka Consumer - Requests	kafka.consumer.consumer-metrics.request-rate	The average number of requests sent per second for a node.	PerSecond
Kafka Consumer - Request size	kafka.consumer.consumer-metrics.request-size-avg	The average size of all requests in the window.	Byte
Kafka Consumer - Incoming byte rate	kafka.consumer.consumer-metrics.incoming-byte-rate	Bytes/second read off all sockets.	BytePerSecond
Kafka Consumer - Outgoing byte rate	kafka.consumer.consumer-metrics.outgoing-byte-rate	The average number of outgoing bytes sent per second to all servers.	BytePerSecond
Kafka Consumer - Request latency	kafka.consumer.consumer-metrics.request-latency-avg	The average request latency in ms for a node.	MilliSecond
Kafka Consumer - Messages consumed rate	kafka.consumer.consumer-metrics.records-consumed-rate	The average number of records consumed per second.	PerSecond
Kafka Consumer - Bytes consumed rate	kafka.consumer.consumer-metrics.bytes-consumed-rate	The average number of bytes consumed per second for a topic.	PerSecond
Kafka Consumer - Fetch latency	kafka.consumer.consumer-metrics.fetch-latency-avg	The average time taken for a fetch request.	MilliSecond
Kafka Consumer - Consumer lag	kafka.consumer.consumer-metrics.records-lag	The latest lag of the partition.	Count
Kafka Consumer - Consumer lag average	kafka.consumer.consumer-metrics.records-lag-avg	The average lag of the partition.	Count
Kafka Consumer - Consumer lag maximum	kafka.consumer.consumer-metrics.records-lag-max	The max lag of the partition.	Count

Metric name	Metric key	Description	Unit
Kafka Server - Purgatory Produce Size	kafka.server.purgatory.produce-delay-size	Requests waiting in the produce purgatory.	Count
Kafka Server - Purgatory Fetch Size	kafka.server.purgatory.fetch-delay-size	Requests waiting in the fetch purgatory.	Count

Metric name	Metric key	Description	Unit
Kafka Producer - Incoming byte rate	kafka.producer.producer-metrics.incoming-byte-rate	The average number of responses received per second for a node.	BytePerSecond
Kafka Producer - Outgoing byte rate	kafka.producer.producer-metrics.outgoing-byte-rate	The average number of outgoing bytes sent per second to all servers.	BytePerSecond
Kafka Producer - I/O Wait time	kafka.producer.producer-metrics.io-wait-time-ns-avg	The average length of time the I/O thread spent waiting for a socket ready for reads or writes in nanoseconds.	NanoSecond
Kafka Producer - Response rate	kafka.producer.producer-metrics.response-rate	Responses received sent per second.	PerSecond
Kafka Producer - Request latency	kafka.producer.producer-metrics.request-latency-avg	The average request latency in ms.	MilliSecond
Kafka Producer - Compression rate	kafka.producer.producer-metrics.compression-rate-avg	The average compression rate of record batches, defined as the average ratio of the compressed batch size over the uncompressed size.	PerSecond
Kafka Producer - Request size	kafka.producer.producer-metrics.request-size-avg	The average size of all requests in the window.	Byte
Kafka Producer - Requests	kafka.producer.producer-metrics.request-rate	The average number of requests sent per second.	PerSecond
Kafka Producer - Byte rate	kafka.producer.producer-topic-metrics.byte-rate	The average number of bytes sent per second for a topic.	BytePerSecond
Kafka Producer - Compression rate	kafka.producer.producer-topic-metrics.compression-rate	The average compression rate of record batches for a topic, defined as the average ratio of the compressed batch size over the uncompressed size.	BytePerSecond
Kafka Producer - Failed Requests Rate	kafka.producer.producer-topic-metrics.record-error-rate	The average per-second number of record sends that resulted in errors for a topic.	PerSecond
Kafka Producer - Requests Sent Rate	kafka.producer.producer-topic-metrics.record-send-rate	The average number of records sent per second for a topic.	PerSecond

Metric name	Metric key	Description	Unit
Kafka Connector - Status	kafka.connector.status	Equals 1 if the status is running, 0 otherwise.	Count
Kafka Connector - Task status	kafka.connector.task.status	Equals 1 if the status is running, 0 otherwise.	Count
Kafka Connector - Task pause ratio	kafka.connector.task.pause-ratio	The fraction of time this task has spent in the pause state.	Count
Kafka Connector - Task running ratio	kafka.connector.task.running-ratio	The fraction of time this task has spent in the running state.	Count
Kafka Connector - Task success ratio	kafka.connector.task.offset-commit-success-percentage	The average percentage of this task's offset commit attempts that succeeded.	Percent
Kafka Connector - Task commit time (max)	kafka.connector.task.offset-commit-max-time-ms	The maximum time in milliseconds taken by this task to commit offsets.	MilliSecond
Kafka Connector - Task failure ratio	kafka.connector.task.offset-commit-failure-percentage	The maximum time in milliseconds taken by this task to commit offsets.	Percent
Kafka Connector - Task commit time (avg)	kafka.connector.task.offset-commit-avg-time-ms	The average time in milliseconds taken by this task to commit offsets.	MilliSecond
Kafka Connector - Task batch size (max)	kafka.connector.task.batch-size-max	The maximum size of the batches processed by the connector.	Byte
Kafka Connector - Task batch size (avg)	kafka.connector.task.batch-size-avg	The average size of the batches processed by the connector.	Byte

Metric name	Metric key	Description	Unit
Kafka Controller - Offline partitions	kafka.controller.KafkaController.OfflinePartitionsCount	The number of partitions that don't have an active leader and are therefore not writable or readable.	Count
Kafka Controller - Active cluster controllers	kafka.controller.KafkaController.ActiveControllerCount.Value	Indicates whether the broker is the controller broker.	Count

Metric name	Metric key	Description	Unit
Kafka Server - Handler Pool Idle Percent Rate	kafka.server.handler.average-idle-percent.rate	The average fraction of time the request handler threads are idle. Values are between 0 meaning all resources are used and 1 meaning all resources are available.	PerSecond

Metric name	Metric key	Description	Unit
Kafka Server - Disk Read Rate	kafka.server.disk.read-bytes	The total number of bytes read by the broker process, including reads from all disks. The total doesn't include reads from page cache. Available only on Linux-based systems.	PerSecond
Kafka Server - Disk Write Rate	kafka.server.disk.write-bytes	The total number of bytes written by the broker process, including writes from all disks. Available only on Linux-based systems.	PerSecond

Metric name	Metric key	Description	Unit
Kafka Broker - Incoming byte rate	kafka.server.BrokerTopicMetrics.BytesInPerSec.OneMinuteRate	The rate at which data sent from producers is consumed by the broker.	BytePerSecond
Kafka Broker - Outgoing byte rate	kafka.server.BrokerTopicMetrics.BytesOutPerSec.OneMinuteRate	The rate at which data sent from other brokers is consumed by the follower broker.	BytePerSecond
Kafka Broker - Messages in rate	kafka.server.BrokerTopicMetrics.MessagesInPerSec.OneMinuteRate	The rate at which individual messages are consumed by the broker.	PerSecond
Kafka Broker - Follower fetch requests rate	kafka.server.BrokerTopicMetrics.TotalFollowerFetchRequestsPerSec.OneMinuteRate	The follower fetch request rate for the broker.	PerSecond
Kafka Broker - Produce message conversions rate	kafka.server.BrokerTopicMetrics.ProduceMessageConversionsPerSec.OneMinuteRate	The rate at which produce messages are converted, by topic.	PerSecond
Kafka Broker - Partitions	kafka.server.ReplicaManager.PartitionCount	The number of partitions in the broker.	Count
Kafka Broker - Under replicated partitions	kafka.server.ReplicaManager.UnderReplicatedPartitions	The number of partitions that have not been fully replicated in the follower replicas.	Count
Kafka Broker - Produce request rate	kafka.server.BrokerTopicMetrics.TotalProduceRequestsPerSec.OneMinuteRate	The produce request rate per second.	PerSecond
Kafka Broker - Fetch request rate	kafka.server.BrokerTopicMetrics.TotalFetchRequestsPerSec.OneMinuteRate	The fetch request rate per second.	PerSecond
Kafka Broker - Failed produce requests	kafka.server.BrokerTopicMetrics.FailedProduceRequestsPerSec.OneMinuteRate	The produce request rate for requests that failed.	PerSecond
Kafka Broker - Failed fetch requests	kafka.server.BrokerTopicMetrics.FailedFetchRequestsPerSec.OneMinuteRate	The fetch request rate for requests that failed.	PerSecond
Kafka Server - Max follower lag	kafka.server.ReplicaFetcherManager.MaxLag.Replica.Value	The maximum lag between the time that messages are received by the leader replica and by the follower replicas.	Count
Kafka Server - Current follower lag	kafka.server.FetcherLagMetrics.ConsumerLag.Value	The lag in number of messages per follower replica.	Count
Kafka Server - Fetch Conversions Rate	kafka.server.FetchConversionsRate.OneMinuteRate	-	Count
Kafka Server - Produce Conversions Rate	kafka.server.ProduceConversionsRate.OneMinuteRate	-	Count

Metric name	Metric key	Description	Unit
Kafka Controller - Leader election rate	kafka.controller.ControllerStats.LeaderElectionRateAndTimeMs.OneMinuteRate	The broker leader election rate and latency in milliseconds. This is non-zero when there are broker failures.	MilliSecond
Kafka Controller - Unclean election rate	kafka.controller.ControllerStats.UncleanLeaderElectionsPerSec.OneMinuteRate	The unclean broker leader election rate. Should be 0.	PerSecond
Kafka Server - Leader count	kafka.server.ReplicaManager.LeaderCount.Value	The number of replicas for which this broker is the leader.	Count

Metric name	Metric key	Description	Unit
Kafka Network - Produce requests per second	kafka.network.RequestMetrics.RequestsPerSec.Produce.OneMinuteRate	The total number of requests made for produce per second	PerSecond
Kafka Network - FetchConsumer requests per second	kafka.network.RequestMetrics.RequestsPerSec.FetchConsumer.OneMinuteRate	The total number of requests made for fetch consumer per second	PerSecond
Kafka Network - FetchFollower requests per second	kafka.network.RequestMetrics.RequestsPerSec.FetchFollower.OneMinuteRate	The total number of requests made for fetch follower per second	PerSecond
Kafka Network - Total time per Produce request	kafka.network.RequestMetrics.TotalTimeMs.Produce.Count	Total time, in milliseconds, spent processing requests, for produce.	MilliSecond
Kafka Network - Total time per FetchConsumer request	kafka.network.RequestMetrics.TotalTimeMs.FetchConsumer.Count	Total time, in milliseconds, spent processing requests, for fetch consumer.	MilliSecond
Kafka Network - Total time per FetchFollower request	kafka.network.RequestMetrics.TotalTimeMs.FetchFollower.Count	Total time, in milliseconds, spent processing requests, for pfetch follower.	MilliSecond
Kafka Network - Request queue size	kafka.network.RequestChannel.RequestQueueSize.Value	Size of the request queue.	Count

Metric name	Metric key	Description	Unit
Kafka Server - ZooKeeper disconnects	kafka.server.SessionExpireListener.ZooKeeperDisconnectsPerSec.OneMinuteRate	A meter that provides the number of recent ZooKeeper client disconnects.	PerSecond
Kafka Server - ZooKeeper expires	kafka.server.SessionExpireListener.ZooKeeperExpiresPerSec.OneMinuteRate	The number of ZooKeeper sessions that have expired.	PerSecond
Kafka Server - Zookeeper Active Connections	kafka.server.active-connections	The number of currently open connections to the broker.	Count

Metric name	Metric key	Description	Unit
Kafka Connect - Requests	kafka.connect.connect-metrics.request-rate	The average number of requests sent per second.	PerSecond
Kafka Connect - Outgoing byte rate	kafka.connect.connect-metrics.outgoing-byte-rate	The average number of outgoing bytes sent per second to all servers.	BytePerSecond
Kafka Connect - Request size	kafka.connect.connect-metrics.request-size-avg	The average size of all requests in the window.	Byte
Kafka Connect - Incoming byte rate	kafka.connect.connect-metrics.incoming-byte-rate	Bytes/second read off all sockets.	BytePerSecond

Metric name	Metric key	Description	Unit
Kafka Log - Log flush 95th percentile	kafka.log.LogFlushStats.LogFlushRateAndTimeMs.Percentile95th	Log flush rate and time in milliseconds.	MilliSecond
Kafka Log - Log flush mean time	kafka.log.LogFlushStats.LogFlushRateAndTimeMs.Mean	Log flush rate and time in milliseconds.	MilliSecond

Full version history

To have more information on how to install the downloaded package, please follow the instructions on this page.

ReleaseDate

Full version history

🚀 Improved in this version:

[🔴 Breaking change] Refactored entity creation for consumers and producers. The dimension client-id is no longer used to generate these entities but only the process_group_instanceand the host will be considered. This will generate new entity IDs for all your existing consumers and producers, which will break any configuration (metric events, dashboards, etc.) that use those individual IDs. The name of these entities will also change, as it was based on the client-idpreviously.
Charts in consumers and producers have been changed to show metrics per client-id, instead of per entity.
Added descriptions to all extension metrics

✨ New in this version:

Added 8 new metrics related to the Kafka Connector.

🐛 Fixed in this version:

As client-id is no longer used for generating consumer and producer entities, we no longer run into the entity dimension limitation that prevented new data from being ingested.

Full version history

Broker now uses the hostname instead of the host ID, when available.
Entity lists for broker, producer and consumer now show the host and the process they are related to, as well as allowing you to filter by those entities.
Removed any reference to zookeeper metrics or entities. To get insight into Zookeeper, activate the recently released standalone extension.

Full version history

Changed the name of the out-of-the-box alerts to include the word Kafka in the title for increased accessibility.
Added the following metrics under the zookeeper-metrics feature set:
- kafka.zookeeper.server.maxRequestLatency
- kafka.zookeeper.server.minRequestLatency
- kafka.zookeeper.server.avgRequestLatency
- kafka.zookeeper.server.aliveConnections
- kafka.zookeeper.server.packetsReceived.count
- kafka.zookeeper.server.packetsSent.count
- kafka.zookeeper.server.outstandingRequests
Added custom topology for the zookeeper
Added UA screens for the zookeeper
Modified the out-of-the-box dashboard to include data about the zookeeper

Full version history

Added a new metric kafka.consumer.consumer-metrics.records-lag to show latest consumer lag instead of the already available consumer lag average and maximum.
Removed the status dimension from the kafka.connector.status and kafka.connector.task.status as the framework is not ready to report such dimensions. The metrics now report 1 when the status is running or 0 otherwise.

Full version history

Fixed a bug where consumer-fetch-manager-metrics where not being captured correctly.
Added connector task status to metrics.
Replace entity lists for Host and Process Group Instance with properties in broker, consumer and producer screens.

Full version history

Fixed a bug where metrics for kafka.consumer:type=consumer-fetch-manager-metrics where not being populated:
- kafka.consumer.consumer-metrics.records-lag-avg
- kafka.consumer.consumer-metrics.records-lag-max
- kafka.consumer.consumer-metrics.records-consumed-rate
- kafka.consumer.consumer-metrics.bytes-consumed-rate
- kafka.consumer.consumer-metrics.fetch-latency-avg

Full version history

Add topic dimension to existing metrics:
- kafka.server.BrokerTopicMetrics.BytesInPerSec.OneMinuteRate
- kafka.server.BrokerTopicMetrics.BytesOutPerSec.OneMinuteRate
- kafka.server.BrokerTopicMetrics.MessagesInPerSec.OneMinuteRate
- kafka.server.BrokerTopicMetrics.TotalProduceRequestsPerSec.OneMinuteRate
Add new metrics:
- kafka.server.BrokerTopicMetrics.TotalFollowerFetchRequestsPerSec.OneMinuteRate
- kafka.server.BrokerTopicMetrics.ProduceMessageConversionsPerSec.OneMinuteRate
Modified UA screens to include the metrics and dimensions listed above:
- Kafka Broker screen
- Kafka Topic screen
- Host injected tile
- Process Group Instance injected tile

Full version history

Kafka Server - Current follower lag: Fixed typo in metric key, changed from kafka.server.FetcherLagMetrics.CosumerLag.Value to kafka.server.FetcherLagMetrics.ConsumerLag.Value
- This is a breaking change if you're using this metric in dashboards, metric events or other configurations. You will need to reference the new key.
Added connector status metric
Added primary entity for most metrics
Added navigation card to all screens

Full version history

No release notes

All

1 Results filtered by:

Tricentis NeoLoad

Automated Load Testing tool enabling you to run Continuous Performance Testing.

Reach out to certified Dynatrace partners to solve your unique use-case

Spica Solutions

Certified individuals: 30Endorsements: Services Endorsed Partner

Authorized Sales Partner

Alanata

Certified individuals: 30Endorsements: Services Endorsed Partner

Premier Sales Partner

avodaq AG

Certified individuals: 31Endorsements: Services Endorsed Partner

Advanced Sales Partner

Moviri

Certified individuals: 14

Advanced Sales Partner

AskMe Solutions & Consultants Co Ltd

Certified individuals: 30Endorsements: Services Endorsed Partner

Authorized Sales Partner

Eviden

Certified individuals: 79Endorsements: Services Endorsed Partner

Premier Sales Partner

Omnilogy

Certified individuals: 38Endorsements: Services Endorsed Partner

Premier Sales Partner

Matrix

Certified individuals: 14

Premier Sales Partner

Phenisys

Certified individuals: 32Endorsements: Services Endorsed Partner

Premier Sales Partner

Spindox

Certified individuals: 11

Authorized Sales Partner