Confluent Platform Demo (cp-demo)

This example builds a full Confluent Platform deployment with an Apache Kafka® event streaming application using ksqlDB and Kafka Streams for stream processing, and all the components have security enabled end-to-end. Follow the accompanying guided tutorial that steps through the example so that you can learn how it all works together.

Overview

Use Case

The use case is an Apache Kafka® event streaming application that processes real-time edits to real Wikipedia pages.

image

Wikimedia Foundation has IRC channels that publish edits happening to real wiki pages (e.g. #en.wikipedia, #en.wiktionary) in real time. Using Kafka Connect, a Kafka source connector kafka-connect-irc streams raw messages from these IRC channels, and a custom Kafka Connect transform kafka-connect-transform-wikiedit transforms these messages and then the messages are written to a Kafka cluster. This example uses ksqlDB and a Kafka Streams application for data processing. Then a Kafka sink connector kafka-connect-elasticsearch streams the data out of Kafka, and the data is materialized into Elasticsearch for analysis by Kibana. Confluent Replicator is also copying messages from a topic to another topic in the same cluster. All data is using Confluent Schema Registry and Avro. Confluent Control Center is managing and monitoring the deployment.

Data Pattern

Data pattern is as follows:

Components Consumes From Produces To
IRC source connector Wikipedia wikipedia.parsed
ksqlDB wikipedia.parsed ksqlDB streams and tables
Kafka Streams application wikipedia.parsed wikipedia.parsed.count-by-channel
Confluent Replicator wikipedia.parsed wikipedia.parsed.replica
Elasticsearch sink connector WIKIPEDIABOT (from ksqlDB) Elasticsearch/Kibana

Run Example

Prerequisites

This example has been validated with:

  • Docker version 17.06.1-ce
  • Docker Compose version 1.14.0 with Docker Compose file format 2.3
  • Java version 1.8.0_92
  • MacOS 10.15.3 (note for Ubuntu environments)
  • OpenSSL 1.1.1d
  • git
  • jq

Docker

This is a Docker environment and has all services running on one host. It is meant exclusively to easily demo Confluent Platform, but in production, do not deploy all Confluent Platform services on a single host.

Also, in production, Confluent Control Center should be deployed with a valid license and with its own dedicated metrics cluster, separate from the cluster with production data. Using a dedicated metrics cluster is more resilient because it continues to provide system health monitoring even if the production traffic cluster experiences issues.

If you prefer non-Docker examples, please go to confluentinc/examples GitHub repository.

Start Example

  1. In Docker’s advanced settings, increase the memory dedicated to Docker to at least 8GB (default is 2GB).

  2. Clone the confluentinc/cp-demo GitHub repository:

    git clone https://github.com/confluentinc/cp-demo
    
  3. Navigate to the cp-demo directory and switch to the Confluent Platform release branch:

    cd cp-demo
    git checkout 6.0.0-post
    
  4. From the cp-demo directory, start the entire example by running a single command that generates the keys and certificates, brings up the Docker containers, and configures and validates the environment. This takes approximately 7 minutes to complete.

    ./scripts/start.sh
    
  5. Using a web browser, view the Confluent Control Center GUI at http://localhost:9021. For this tutorial, log in as superUser and password superUser, which has super user access to the cluster. You may also log in as other users to learn how each user’s view changes depending on their permissions.

  6. To see the end of the entire pipeline, view the Kibana dashboard at http://localhost:5601/app/kibana#/dashboard/Wikipedia

  7. After you run through the guided tutorial below, apply the concepts you learn here to build your own event streaming pipeline in Confluent Cloud, a fully managed, cloud-native event streaming platform powered by Kafka. When you sign up for Confluent Cloud, use the promo code C50INTEG to receive an additional $50 free usage (details).

Guided Tutorial

Brokers

  1. Select the cluster named “Kafka Raleigh”.

    ../../../_images/cluster_raleigh.png
  2. Click on “Brokers”.

  3. View the status of the Brokers in the cluster:

    ../../../_images/landing_page.png
  4. Click through on Production or Consumption to view: Production and Consumption metrics, Broker uptime, Partitions: online, under replicated, total replicas, out of sync replicas, Disk utilization, System: network pool usage, request pool usage.

    ../../../_images/broker_metrics.png

Topics

  1. Confluent Control Center can manage topics in a Kafka cluster. Click on “Topics”.

  2. Scroll down and click on the topic wikipedia.parsed.

    image
  3. View an overview of this topic:

    • Throughput
    • Partition replication status
    image
  4. View which brokers are leaders for which partitions and where all partitions reside.

    image
  5. Inspect messages for this topic, in real-time.

    image
  6. View the schema for this topic. For wikipedia.parsed, the topic value is using a Schema registered with Schema Registry (the topic key is just a string).

    image
  7. View configuration settings for this topic.

    image
  8. Return to “All Topics”, click on wikipedia.parsed.count-by-channel to view the output topic from the Kafka Streams application.

    image
  9. Return to the All topics view and click the + Add a topic button on the top right to create a new topic in your Kafka cluster. You can also view and edit settings of Kafka topics in the cluster. Read more on Confluent Control Center topic management.

    image

Kafka Connect

This example runs three connectors:

  • IRC source connector
  • Elasticsearch sink connector
  • Confluent Replicator

They are running on a Connect worker that is configured with Confluent Platform security features. The Connect worker’s embedded producer is configured to be idempotent, exactly-once in order semantics per partition (in the event of an error that causes a producer retry, the same message—which is still sent by the producer multiple times—will only be written to the Kafka log on the broker once).

  1. Confluent Control Center uses the Kafka Connect API to manage multiple connect clusters. Click on “Connect”.

  2. Select connect1, the name of the cluster of Connect workers.

    ../../../_images/connect_default.png
  3. Verify the connectors running in this example:

    • source connector wikipedia-irc view the example’s IRC source connector configuration file.
    • source connector replicate-topic: view the example’s Replicator connector configuration file.
    • sink connector elasticsearch-ksqldb consuming from the Kafka topic WIKIPEDIABOT: view the example’s Elasticsearch sink connector configuration file.
    ../../../_images/connector_list.png
  4. Click any connector name to view or modify any details of the connector configuration and custom transforms.

    ../../../_images/connect_replicator_settings.png

ksqlDB

In this example, ksqlDB is authenticated and authorized to connect to the secured Kafka cluster, and it is already running queries as defined in the ksqlDB command file . Its embedded producer is configured to be idempotent, exactly-once in order semantics per partition (in the event of an error that causes a producer retry, the same message—which is still sent by the producer multiple times—will only be written to the Kafka log on the broker once).

  1. In the navigation bar, click ksqlDB.

  2. From the list of ksqlDB applications, select wikipedia.

    image
  3. View the ksqlDB Flow to see the streams and tables created in the example, and how they relate to one another.

    image
  4. Use Confluent Control Center to interact with ksqlDB, or run ksqlDB CLI to get to the ksqlDB CLI prompt.

    docker-compose exec ksqldb-cli bash -c 'ksql -u ksqlDBUser -p ksqlDBUser http://ksqldb-server:8088'
    
  5. View the existing ksqlDB streams. (If you are using the ksqlDB CLI, at the ksql> prompt type SHOW STREAMS;)

    image
  6. Click on WIKIPEDIA to describe the schema (fields or columns) of an existing ksqlDB stream. (If you are using the ksqlDB CLI, at the ksql> prompt type DESCRIBE WIKIPEDIA;)

    image
  7. View the existing ksqlDB tables. (If you are using the ksqlDB CLI, at the ksql> prompt type SHOW TABLES;).

    image
  8. View the existing ksqlDB queries, which are continuously running. (If you are using the ksqlDB CLI, at the ksql> prompt type SHOW QUERIES;).

    image
  9. View messages from different ksqlDB streams and tables. Click on your stream of choice and then click Query stream to open the Query Editor. The editor shows a pre-populated query, like select * from WIKIPEDIA EMIT CHANGES;, and it shows results for newly arriving data.

    image
  10. Click ksqlDB Editor and run the SHOW PROPERTIES; statement. You can see the configured ksqlDB server properties and check these values with the docker-compose.yml file.

    image
  11. This example creates two streams EN_WIKIPEDIA_GT_1 and EN_WIKIPEDIA_GT_1_COUNTS, and the reason is to demonstrate how ksqlDB windows work. EN_WIKIPEDIA_GT_1 counts occurences with a tumbling window, and for a given key it writes a null into the table on the first seen message. The underlying Kafka topic for EN_WIKIPEDIA_GT_1 does not filter out those nulls, but to send just the counts greater than one downstream, there is a separate Kafka topic for ``EN_WIKIPEDIA_GT_1_COUNTS which does filter out those nulls (e.g., the query has a clause where ROWTIME is not null). From the bash prompt, view those underlying Kafka topics.

  • View messages in the topic EN_WIKIPEDIA_GT_1 (jump to offset 0/partition 0), and notice the nulls:

    image
  • For comparison, view messages in the topic EN_WIKIPEDIA_GT_1_COUNTS (jump to offset 0/partition 0), and notice no nulls:

    image
  1. The ksqlDB processing log captures per-record errors during processing to help developers debug their ksqlDB queries. In this example, the processing log uses mutual TLS (mTLS) authentication, as configured in the custom log4j properties file, to write entries into a Kafka topic. To see it in action, in the ksqlDB editor run the following “bad” query for 20 seconds:
SELECT ucase(cast(null as varchar)) FROM wikipedia EMIT CHANGES;

No records should be returned from this query. ksqlDB writes errors into the processing log for each record. View the processing log topic ksql-clusterksql_processing_log with topic inspection (jump to offset 0/partition 0) or the corresponding ksqlDB stream KSQL_PROCESSING_LOG with the ksqlDB editor (set auto.offset.reset=earliest).

SELECT * FROM KSQL_PROCESSING_LOG EMIT CHANGES;

Consumers

  1. Confluent Control Center enables you to monitor consumer lag and throughput performance. Consumer lag is the topic’s high water mark (latest offset for the topic that has been written) minus the current consumer offset (latest offset read for that topic by that consumer group). Keep in mind the topic’s write rate and consumer group’s read rate when you consider the significance the consumer lag’s size. Click on “Consumers”.

  2. Consumer lag is available on a per-consumer basis, including the embedded Connect consumers for sink connectors (e.g., connect-elasticsearch-ksqldb), ksqlDB queries (e.g., consumer groups whose names start with _confluent-ksql-default_query_), console consumers (e.g., WIKIPEDIANOBOT-consumer), etc. Consumer lag is also available on a per-topic basis.

    image
  3. View consumer lag for the persistent ksqlDB “Create Stream As Select” query CSAS_WIKIPEDIABOT, which is displayed as _confluent-ksql-ksql-clusterquery_CSAS_WIKIPEDIABOT_5 in the consumer group list.

    image
  4. View consumer lag for the Kafka Streams application under the consumer group id wikipedia-activity-monitor. This application is run by the cnfldemos/cp-demo-kstreams Docker container (application source code).

    image
  5. Consumption metrics are available on a per-consumer basis. These consumption charts are only populated if Confluent Monitoring Interceptors are configured, as they are in this example. You can view % messages consumed and end-to-end latency. View consumption metrics for the persistent ksqlDB “Create Stream As Select” query CSAS_WIKIPEDIABOT, which is displayed as _confluent-ksql-default_query_CSAS_WIKIPEDIABOT_0 in the consumer group list.

    image
  6. Confluent Control Center shows which consumers in a consumer group are consuming from which partitions and on which brokers those partitions reside. Confluent Control Center updates as consumer rebalances occur in a consumer group. Start consuming from topic wikipedia.parsed with a new consumer group app with one consumer consumer_app_1. It runs in the background.

    ./scripts/app/start_consumer_app.sh 1
    
  7. Let this consumer group run for 2 minutes until Confluent Control Center shows the consumer group app with steady consumption. This consumer group app has a single consumer consumer_app_1 consuming all of the partitions in the topic wikipedia.parsed.

    image
  8. Add a second consumer consumer_app_2 to the existing consumer group app.

    ./scripts/app/start_consumer_app.sh 2
    
  9. Let this consumer group run for 2 minutes until Confluent Control Center shows the consumer group app with steady consumption. Notice that the consumers consumer_app_1 and consumer_app_2 now share consumption of the partitions in the topic wikipedia.parsed.

    image
  10. From the Brokers -> Consumption view, click on a point in the Request latency line graph to view a breakdown of latencies through the entire request lifecycle.

    image

Confluent Replicator

Confluent Replicator copies data from a source Kafka cluster to a destination Kafka cluster. The source and destination clusters are typically different clusters, but in this example, Replicator is doing intra-cluster replication, i.e., the source and destination Kafka clusters are the same. As with the rest of the components in the solution, Confluent Replicator is also configured with security.

  1. View Replicator status and throughput in a dedicated view in Confluent Control Center.

    image
  2. Consumers: monitor throughput and latency of Confluent Replicator. Replicator is a Kafka Connect source connector and has a corresponding consumer group connect-replicator.

    image
  3. View Replicator Consumer Lag.

    image
  4. View Replicator Consumption metrics.

    image
  5. Connect: pause the Replicator connector in Settings by pressing the pause icon in the top right and wait for 10 seconds until it takes effect. This stops consumption for the related consumer group.

    image
  6. Observe that the connect-replicator consumer group has stopped consumption.

    ../../../_images/replicator_stopped.png
  7. Restart the Replicator connector.

  8. Observe that the connect-replicator consumer group has resumed consumption. Notice several things:

    • Even though the consumer group connect-replicator was not running for some of this time, all messages are shown as delivered. This is because all bars are time windows relative to produce timestamp.
    • The latency peaks and then gradually decreases, because this is also relative to the produce timestamp.
  9. Next step: Learn more about Replicator with the Replicator Tutorial.

Security

All the Confluent Platform components and clients in this example are enabled with many security features.

You can see each component’s security configuration in the example’s docker-compose.yml file.

Note

This example showcases a secure Confluent Platform for educational purposes and is not meant to be complete best practices. There are certain differences between what is shown in the example and what you should do in production:

  • Authorize users only for operations that they need, instead of making all of them super users
  • If the PLAINTEXT security protocol is used, these ANONYMOUS usernames should not be configured as super users
  • Consider not even opening the PLAINTEXT port if SSL or SASL_SSL are configured

There is an OpenLDAP server running in the example, and each Kafka broker in the demo is configured with Metadata Service (MDS) and can talk to LDAP so that it can authenticate clients and Confluent Platform services and clients.

ZooKeeper has two listener ports:

Name Protocol In this example, used for … ZooKeeper
N/A SASL/DIGEST-MD5 Validating trial license for REST Proxy and Schema Registry. (no TLS support) 2181
N/A mTLS Broker communication (kafka1, kafka2) 2182

Each broker has five listener ports:

Name Protocol In this example, used for … kafka1 kafka2
N/A MDS Authorization via RBAC 8091 8092
INTERNAL SASL_PLAINTEXT CP Kafka clients (e.g. Confluent Metrics Reporter), SASL_PLAINTEXT 9091 9092
TOKEN SASL_SSL Confluent Platform service (e.g. Schema Registry) when they need to use impersonation 10091 10092
SSL SSL End clients, (e.g. stream-demo), with SSL no SASL 11091 11092
CLEAR PLAINTEXT No security, available as a backdoor; for demo and learning only 12091 12092

End clients (non-CP clients):

  • Authenticate using mTLS via the broker SSL listener.
  • If they are also using Schema Registry, authenticate to Schema Registry via LDAP.
  • If they are also using Confluent Monitoring interceptors, authenticate using mTLS via the broker SSL listener.
  • Should never use the TOKEN listener which is meant only for internal communication between Confluent components.
  • See client configuration used in the example by the streams-demo container running the Kafka Streams application wikipedia-activity-monitor.
  1. Verify the ports on which the Kafka brokers are listening with the following command, and they should match the table shown below:

    docker-compose logs kafka1 | grep "Registered broker 1"
    docker-compose logs kafka2 | grep "Registered broker 2"
    
  2. For example only: Communicate with brokers via the PLAINTEXT port, client security configurations are not required

    # CLEAR/PLAINTEXT port
    docker-compose exec kafka1 kafka-consumer-groups \
       --list \
       --bootstrap-server kafka1:12091
    
  3. End clients: Communicate with brokers via the SSL port, and SSL parameters configured via the --command-config argument for command line tools or --consumer.config for kafka-console-consumer.

    # SSL/SSL port
    docker-compose exec kafka1 kafka-consumer-groups \
       --list \
       --bootstrap-server kafka1:11091 \
       --command-config /etc/kafka/secrets/client_without_interceptors_ssl.config
    
  4. If a client tries to communicate with brokers via the SSL port but does not specify the SSL parameters, it fails

    # SSL/SSL port
    docker-compose exec kafka1 kafka-consumer-groups \
       --list \
       --bootstrap-server kafka1:11091
    

    Your output should resemble:

    ERROR Uncaught exception in thread 'kafka-admin-client-thread | adminclient-1': (org.apache.kafka.common.utils.KafkaThread)
    java.lang.OutOfMemoryError: Java heap space
    ...
    
  5. Communicate with brokers via the SASL_PLAINTEXT port, and SASL_PLAINTEXT parameters configured via the --command-config argument for command line tools or --consumer.config for kafka-console-consumer.

    # INTERNAL/SASL_PLAIN port
    docker-compose exec kafka1 kafka-consumer-groups \
       --list \
       --bootstrap-server kafka1:9091 \
       --command-config /etc/kafka/secrets/client_sasl_plain.config
    
  6. Verify which users are configured to be super users.

    docker-compose logs kafka1 | grep "super.users ="
    

    Your output should resemble the following. Notice this authorizes each service name which authenticates as itself, as well as the unauthenticated PLAINTEXT which authenticates as ANONYMOUS (for demo purposes only):

    kafka1            |    super.users = User:admin;User:mds;User:superUser;User:ANONYMOUS
    
  7. Verify that LDAP user appSA (which is not a super user) can consume messages from topic wikipedia.parsed. Notice that it is configured to authenticate to brokers with mTLS and authenticate to Schema Registry with LDAP.

    docker-compose exec connect kafka-avro-console-consumer \
      --bootstrap-server kafka1:11091,kafka2:11092 \
      --consumer-property security.protocol=SSL \
      --consumer-property ssl.truststore.location=/etc/kafka/secrets/kafka.appSA.truststore.jks \
      --consumer-property ssl.truststore.password=confluent \
      --consumer-property ssl.keystore.location=/etc/kafka/secrets/kafka.appSA.keystore.jks \
      --consumer-property ssl.keystore.password=confluent \
      --consumer-property ssl.key.password=confluent \
      --property schema.registry.url=https://schemaregistry:8085 \
      --property schema.registry.ssl.truststore.location=/etc/kafka/secrets/kafka.appSA.truststore.jks \
      --property schema.registry.ssl.truststore.password=confluent \
      --property basic.auth.credentials.source=USER_INFO \
      --property schema.registry.basic.auth.user.info=appSA:appSA \
      --group wikipedia.test \
      --topic wikipedia.parsed \
      --max-messages 5
    
  8. Verify that LDAP user badapp cannot consume messages from topic wikipedia.parsed.

    docker-compose exec connect kafka-avro-console-consumer \
      --bootstrap-server kafka1:11091,kafka2:11092 \
      --consumer-property security.protocol=SSL \
      --consumer-property ssl.truststore.location=/etc/kafka/secrets/kafka.badapp.truststore.jks \
      --consumer-property ssl.truststore.password=confluent \
      --consumer-property ssl.keystore.location=/etc/kafka/secrets/kafka.badapp.keystore.jks \
      --consumer-property ssl.keystore.password=confluent \
      --consumer-property ssl.key.password=confluent \
      --property schema.registry.url=https://schemaregistry:8085 \
      --property schema.registry.ssl.truststore.location=/etc/kafka/secrets/kafka.badapp.truststore.jks \
      --property schema.registry.ssl.truststore.password=confluent \
      --property basic.auth.credentials.source=USER_INFO \
      --property schema.registry.basic.auth.user.info=badapp:badapp \
      --group wikipedia.test \
      --topic wikipedia.parsed \
      --max-messages 5
    

    Your output should resemble:

    ERROR [Consumer clientId=consumer-wikipedia.test-1, groupId=wikipedia.test] Topic authorization failed for topics [wikipedia.parsed]
    org.apache.kafka.common.errors.TopicAuthorizationException: Not authorized to access topics: [wikipedia.parsed]
    
  9. Create role bindings to permit badapp client to consume from topic wikipedia.parsed and its related subject in Schema Registry.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Create the role bindings:

    # Create the role binding for the topic ``wikipedia.parsed``
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:badapp \
        --role ResourceOwner \
        --resource Topic:wikipedia.parsed \
        --kafka-cluster-id $KAFKA_CLUSTER_ID"
    
    # Create the role binding for the group ``wikipedia.test``
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:badapp \
        --role ResourceOwner \
        --resource Group:wikipedia.test \
        --kafka-cluster-id $KAFKA_CLUSTER_ID"
    
    # Create the role binding for the subject ``wikipedia.parsed-value``, i.e., the topic-value (versus the topic-key)
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:badapp \
        --role ResourceOwner \
        --resource Subject:wikipedia.parsed-value \
        --kafka-cluster-id $KAFKA_CLUSTER_ID \
        --schema-registry-cluster-id schema-registry"
    
  10. Verify that LDAP user badapp now can consume messages from topic wikipedia.parsed.

    docker-compose exec connect kafka-avro-console-consumer \
      --bootstrap-server kafka1:11091,kafka2:11092 \
      --consumer-property security.protocol=SSL \
      --consumer-property ssl.truststore.location=/etc/kafka/secrets/kafka.badapp.truststore.jks \
      --consumer-property ssl.truststore.password=confluent \
      --consumer-property ssl.keystore.location=/etc/kafka/secrets/kafka.badapp.keystore.jks \
      --consumer-property ssl.keystore.password=confluent \
      --consumer-property ssl.key.password=confluent \
      --property schema.registry.url=https://schemaregistry:8085 \
      --property schema.registry.ssl.truststore.location=/etc/kafka/secrets/kafka.badapp.truststore.jks \
      --property schema.registry.ssl.truststore.password=confluent \
      --property basic.auth.credentials.source=USER_INFO \
      --property schema.registry.basic.auth.user.info=badapp:badapp \
      --group wikipedia.test \
      --topic wikipedia.parsed \
      --max-messages 5
    
  11. View all the role bindings that were configured for RBAC in this cluster.

    cd scripts/validate
    ./validate_bindings.sh
    
  12. Because ZooKeeper is configured for SASL/DIGEST-MD5, any commands that communicate with ZooKeeper need properties set for ZooKeeper authentication. This authentication configuration is provided by the KAFKA_OPTS setting on the brokers. For example, notice that the throttle script runs on the Docker container kafka1 which has the appropriate KAFKA_OPTS setting. The command would otherwise fail if run on any other container aside from kafka1 or kafka2.

  13. Next step: Learn more about security with the Security Tutorial.

Data Governance with Schema Registry

All the applications and connectors used in this example are configured to automatically read and write Avro-formatted data, leveraging the Confluent Schema Registry .

The security in place between Schema Registry and the end clients, e.g. appSA, is as follows:

  • Encryption: TLS, e.g. client has schema.registry.ssl.truststore.* configurations
  • Authentication: bearer token authentication from HTTP basic auth headers, e.g. client has schema.registry.basic.auth.user.info and basic.auth.credentials.source configurations
  • Authorization: Schema Registry uses the bearer token with RBAC to authorize the client
  1. View the Schema Registry subjects for topics that have registered schemas for their keys and/or values. Notice the curl arguments include (a) TLS information required to interact with Schema Registry which is listening for HTTPS on port 8085, and (b) authentication credentials required for RBAC (using superUser:superUser to see all of them).

    docker-compose exec schemaregistry curl -X GET \
       --cert /etc/kafka/secrets/schemaregistry.certificate.pem \
       --key /etc/kafka/secrets/schemaregistry.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u superUser:superUser \
       https://schemaregistry:8085/subjects | jq .
    

    Your output should resemble:

    [
      "wikipedia.parsed.replica-value",
      "EN_WIKIPEDIA_GT_1_COUNTS-value",
      "WIKIPEDIABOT-value",
      "EN_WIKIPEDIA_GT_1-value",
      "_confluent-ksql-ksql-clusterquery_CTAS_EN_WIKIPEDIA_GT_1_7-Aggregate-Aggregate-Materialize-changelog-value",
      "WIKIPEDIANOBOT-value",
      "_confluent-ksql-ksql-clusterquery_CTAS_EN_WIKIPEDIA_GT_1_7-Aggregate-GroupBy-repartition-value",
      "wikipedia.parsed-value"
    ]
    
  2. Instead of using the superUser credentials, now use client credentials noexist:noexist (user does not exist in LDAP) to try to register a new Avro schema (a record with two fields username and userid) into Schema Registry for the value of a new topic users. It should fail due to an authorization error.

    docker-compose exec schemaregistry curl -X POST \
       -H "Content-Type: application/vnd.schemaregistry.v1+json" \
       --cert /etc/kafka/secrets/schemaregistry.certificate.pem \
       --key /etc/kafka/secrets/schemaregistry.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{ "schema": "[ { \"type\":\"record\", \"name\":\"user\", \"fields\": [ {\"name\":\"userid\",\"type\":\"long\"}, {\"name\":\"username\",\"type\":\"string\"} ]} ]" }' \
       -u noexist:noexist \
       https://schemaregistry:8085/subjects/users-value/versions
    

    Your output should resemble:

    {"error_code":401,"message":"Unauthorized"}
    
  3. Instead of using credentials for a user that does not exist, now use the client credentials appSA:appSA (the user appSA exists in LDAP) to try to register a new Avro schema (a record with two fields username and userid) into Schema Registry for the value of a new topic users. It should fail due to an authorization error, with a different message than above.

    docker-compose exec schemaregistry curl -X POST \
       -H "Content-Type: application/vnd.schemaregistry.v1+json" \
       --cert /etc/kafka/secrets/schemaregistry.certificate.pem \
       --key /etc/kafka/secrets/schemaregistry.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{ "schema": "[ { \"type\":\"record\", \"name\":\"user\", \"fields\": [ {\"name\":\"userid\",\"type\":\"long\"}, {\"name\":\"username\",\"type\":\"string\"} ]} ]" }' \
       -u appSA:appSA \
       https://schemaregistry:8085/subjects/users-value/versions
    

    Your output should resemble:

    {"error_code":40403,"message":"User is denied operation Write on Subject: users-value"}
    
  4. Create a role binding for the appSA client permitting it access to Schema Registry.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Create the role binding:

    # Create the role binding for the subject ``users-value``, i.e., the topic-value (versus the topic-key)
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:appSA \
        --role ResourceOwner \
        --resource Subject:users-value \
        --kafka-cluster-id $KAFKA_CLUSTER_ID \
        --schema-registry-cluster-id schema-registry"
    
  5. Again try to register the schema. It should pass this time. Note the schema id that it returns, e.g. below schema id is 11.

    docker-compose exec schemaregistry curl -X POST \
       -H "Content-Type: application/vnd.schemaregistry.v1+json" \
       --cert /etc/kafka/secrets/schemaregistry.certificate.pem \
       --key /etc/kafka/secrets/schemaregistry.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{ "schema": "[ { \"type\":\"record\", \"name\":\"user\", \"fields\": [ {\"name\":\"userid\",\"type\":\"long\"}, {\"name\":\"username\",\"type\":\"string\"} ]} ]" }' \
       -u appSA:appSA \
       https://schemaregistry:8085/subjects/users-value/versions
    

    Your output should resemble:

    {"id":11}
    
  6. View the new schema for the subject users-value. From Confluent Control Center, click Topics. Scroll down to and click on the topic users and select “SCHEMA”.

    image

    You may alternatively request the schema via the command line:

    docker-compose exec schemaregistry curl -X GET \
       --cert /etc/kafka/secrets/schemaregistry.certificate.pem \
       --key /etc/kafka/secrets/schemaregistry.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       https://schemaregistry:8085/subjects/users-value/versions/1 | jq .
    

    Your output should resemble:

    {
      "subject": "users-value",
      "version": 1,
      "id": 11,
      "schema": "{\"type\":\"record\",\"name\":\"user\",\"fields\":[{\"name\":\"username\",\"type\":\"string\"},{\"name\":\"userid\",\"type\":\"long\"}]}"
    }
    
  7. Describe the topic users. Notice that it has a special configuration confluent.value.schema.validation=true which enables Schema Validation, a data governance feature in Confluent Server that gives operators a centralized location within the Kafka cluster itself to enforce data format correctness. Enabling Schema Validation allows brokers configured with confluent.schema.registry.url to validate that data produced to the topic is using a valid schema.

    docker-compose exec kafka1 kafka-topics \
       --describe \
       --topic users \
       --bootstrap-server kafka1:9091 \
       --command-config /etc/kafka/secrets/client_sasl_plain.config
    

    Your output should resemble:

    Topic: users      PartitionCount: 2       ReplicationFactor: 2    Configs: confluent.value.schema.validation=true
            Topic: users      Partition: 0    Leader: 1       Replicas: 1,2   Isr: 1,2        Offline:
            Topic: users      Partition: 1    Leader: 2       Replicas: 2,1   Isr: 2,1        Offline:
    
  8. Produce a non-Avro message to this topic using kafka-console-producer, and it results in a failure.

    docker-compose exec connect kafka-console-producer \
         --topic users \
         --broker-list kafka1:11091 \
         --producer-property security.protocol=SSL \
         --producer-property ssl.truststore.location=/etc/kafka/secrets/kafka.appSA.truststore.jks \
         --producer-property ssl.truststore.password=confluent \
         --producer-property ssl.keystore.location=/etc/kafka/secrets/kafka.appSA.keystore.jks \
         --producer-property ssl.keystore.password=confluent \
         --producer-property ssl.key.password=confluent
    

    The error should resemble:

    ERROR Error when sending message to topic users with key: null, value: 5 bytes with error: (org.apache.kafka.clients.producer.internals.ErrorLoggingCallback)
    org.apache.kafka.common.InvalidRecordException: This record has failed the validation on broker and hence be rejected.
    
  9. Describe the topic wikipedia.parsed, which is the topic that the kafka-connect-irc source connector is writing to. Notice that it also has enabled Schema Validation.

    docker-compose exec kafka1 kafka-topics \
       --describe \
       --topic wikipedia.parsed \
       --bootstrap-server kafka1:9091 \
       --command-config /etc/kafka/secrets/client_sasl_plain.config
    
  10. Describe the topic wikipedia.parsed.replica, which is the topic that Replicator has replicated from wikipedia.parsed. Notice that it also has enabled Schema Validation, because Replicator default is topic.config.sync=true (see Replicator documentation).

    docker-compose exec kafka1 kafka-topics \
       --describe \
       --topic wikipedia.parsed.replica \
       --bootstrap-server kafka1:9091 \
       --command-config /etc/kafka/secrets/client_sasl_plain.config
    
  11. Next step: Learn more about Schema Registry with the Schema Registry Tutorial.

Confluent REST Proxy

The Confluent REST Proxy is running for optional client access. This demo showcases Confluent REST Proxy in two modes:

  • Standalone service, listening for HTTPS requests on port 8086
  • Embedded service on the Kafka brokers, listening for HTTPS requests on port 8091 on kafka1 and on port 8092 on kafka2 (these REST Proxy ports are shared with the broker’s Metadata Service (MDS) listener)
  1. Use the standalone REST Proxy to try to produce a message to the topic users, referencing schema id 11. This schema was registered in Schema Registry in the previous section. It should fail due to an authorization error.

    docker-compose exec restproxy curl -X POST \
       -H "Content-Type: application/vnd.kafka.avro.v2+json" \
       -H "Accept: application/vnd.kafka.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{"value_schema_id": 11, "records": [{"value": {"user":{"userid": 1, "username": "Bunny Smith"}}}]}' \
       -u appSA:appSA \
       https://restproxy:8086/topics/users
    

    Your output should resemble:

    {"offsets":[{"partition":null,"offset":null,"error_code":40301,"error":"Not authorized to access topics: [users]"}],"key_schema_id":null,"value_schema_id":11}
    
  2. Create a role binding for the client permitting it produce to the topic users.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Create the role binding:

    # Create the role binding for the topic ``users``
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:appSA \
        --role DeveloperWrite \
        --resource Topic:users \
        --kafka-cluster-id $KAFKA_CLUSTER_ID"
    
  3. Again try to produce a message to the topic users. It should pass this time.

    docker-compose exec restproxy curl -X POST \
       -H "Content-Type: application/vnd.kafka.avro.v2+json" \
       -H "Accept: application/vnd.kafka.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{"value_schema_id": 11, "records": [{"value": {"user":{"userid": 1, "username": "Bunny Smith"}}}]}' \
       -u appSA:appSA \
       https://restproxy:8086/topics/users
    

    Your output should resemble:

    {"offsets":[{"partition":1,"offset":0,"error_code":null,"error":null}],"key_schema_id":null,"value_schema_id":11}
    
  4. Create consumer instance my_avro_consumer.

    docker-compose exec restproxy curl -X POST \
       -H "Content-Type: application/vnd.kafka.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{"name": "my_consumer_instance", "format": "avro", "auto.offset.reset": "earliest"}' \
       -u appSA:appSA \
       https://restproxy:8086/consumers/my_avro_consumer
    

    Your output should resemble:

    {"instance_id":"my_consumer_instance","base_uri":"https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance"}
    
  5. Subscribe my_avro_consumer to the users topic.

    docker-compose exec restproxy curl -X POST \
       -H "Content-Type: application/vnd.kafka.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       --data '{"topics":["users"]}' \
       -u appSA:appSA \
       https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance/subscription
    
  6. Try to consume messages for my_avro_consumer subscriptions. It should fail due to an authorization error.

    docker-compose exec restproxy curl -X GET \
       -H "Accept: application/vnd.kafka.avro.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance/records
    

    Your output should resemble:

    {"error_code":40301,"message":"Not authorized to access group: my_avro_consumer"}
    
  7. Create a role binding for the client permitting it access to the consumer group my_avro_consumer.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Create the role binding:

    # Create the role binding for the group ``my_avro_consumer``
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:appSA \
        --role ResourceOwner \
        --resource Group:my_avro_consumer \
        --kafka-cluster-id $KAFKA_CLUSTER_ID"
    
  8. Again try to consume messages for my_avro_consumer subscriptions. It should fail due to a different authorization error.

    # Note: Issue this command twice due to https://github.com/confluentinc/kafka-rest/issues/432
    docker-compose exec restproxy curl -X GET \
       -H "Accept: application/vnd.kafka.avro.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance/records
    
    docker-compose exec restproxy curl -X GET \
       -H "Accept: application/vnd.kafka.avro.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance/records
    

    Your output should resemble:

    {"error_code":40301,"message":"Not authorized to access topics: [users]"}
    
  9. Create a role binding for the client permitting it access to the topic users.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Create the role binding:

    # Create the role binding for the group my_avro_consumer
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:appSA \
        --role DeveloperRead \
        --resource Topic:users \
        --kafka-cluster-id $KAFKA_CLUSTER_ID"
    
  10. Again try to consume messages for my_avro_consumer subscriptions. It should pass this time.

       # Note: Issue this command twice due to https://github.com/confluentinc/kafka-rest/issues/432
       docker-compose exec restproxy curl -X GET \
          -H "Accept: application/vnd.kafka.avro.v2+json" \
          --cert /etc/kafka/secrets/restproxy.certificate.pem \
          --key /etc/kafka/secrets/restproxy.key \
          --tlsv1.2 \
          --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
          -u appSA:appSA \
          https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance/records
    
       docker-compose exec restproxy curl -X GET \
          -H "Accept: application/vnd.kafka.avro.v2+json" \
          --cert /etc/kafka/secrets/restproxy.certificate.pem \
          --key /etc/kafka/secrets/restproxy.key \
          --tlsv1.2 \
          --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
          -u appSA:appSA \
          https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance/records
    
    Your output should resemble:
    
    [{"topic":"users","key":null,"value":{"userid":1,"username":"Bunny Smith"},"partition":1,"offset":0}]
    
  11. Delete the consumer instance my_avro_consumer.

    docker-compose exec restproxy curl -X DELETE \
       -H "Content-Type: application/vnd.kafka.v2+json" \
       --cert /etc/kafka/secrets/restproxy.certificate.pem \
       --key /etc/kafka/secrets/restproxy.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       https://restproxy:8086/consumers/my_avro_consumer/instances/my_consumer_instance
    
  12. For the next few steps, use the REST Proxy that is embedded on the Kafka brokers. Only REST Proxy API v3 is supported this time. Create a role binding for the client to be granted ResourceOwner role for the topic dev_users.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Create the role binding:

    # Create the role binding for the topic ``dev_users``
    docker-compose exec tools bash -c "confluent iam rolebinding create \
        --principal User:appSA \
        --role ResourceOwner \
        --resource Topic:dev_users \
        --kafka-cluster-id $KAFKA_CLUSTER_ID"
    
  13. Create the topic dev_users with embedded REST Proxy.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Use curl to create the topic:

    docker-compose exec restproxy curl -X POST \
       -H "Content-Type: application/json" \
       -H "accept: application/json" \
       -d "{\"topic_name\":\"dev_users\",\"partitions_count\":64,\"replication_factor\":2,\"configs\":[{\"name\":\"cleanup.policy\",\"value\":\"compact\"},{\"name\":\"compression.type\",\"value\":\"gzip\"}]}" \
       --cert /etc/kafka/secrets/mds.certificate.pem \
       --key /etc/kafka/secrets/mds.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       "https://kafka1:8091/kafka/v3/clusters/${KAFKA_CLUSTER_ID}/topics" | jq
    
  14. List topics with embedded REST Proxy to find the newly created dev_users.

    Get the Kafka cluster ID:

    KAFKA_CLUSTER_ID=$(curl -s https://localhost:8091/v1/metadata/id --tlsv1.2 --cacert scripts/security/snakeoil-ca-1.crt | jq -r ".id")
    

    Use curl to list the topics:

    docker-compose exec restproxy curl -X GET \
       -H "Content-Type: application/json" \
       -H "accept: application/json" \
       --cert /etc/kafka/secrets/mds.certificate.pem \
       --key /etc/kafka/secrets/mds.key \
       --tlsv1.2 \
       --cacert /etc/kafka/secrets/snakeoil-ca-1.crt \
       -u appSA:appSA \
       https://kafka1:8091/kafka/v3/clusters/${KAFKA_CLUSTER_ID}/topics | jq '.data[].topic_name'
    

Failed Broker

To simulate a failed broker, stop the Docker container running one of the two Kafka brokers.

  1. Stop the Docker container running Kafka broker 2.

    docker-compose stop kafka2
    
  2. After a few minutes, observe the Broker summary show that the number of brokers has decreased from 2 to 1, and there are many under replicated partitions.

    image
  3. View Topic information details to see that there are out of sync replicas on broker 2.

    image
  4. Look at the production and consumption metrics and notice that the clients are all still working.

    image
  5. Restart the Docker container running Kafka broker 2.

    docker-compose start kafka2
    
  6. After about a minute, observe the Broker summary in Confluent Control Center. The broker count has recovered to 2, and the topic partitions are back to reporting no under replicated partitions.

    image
  7. Click on the broker count 2 inside the “Brokers” box and when the “Brokers overview” pane appears, click inside the “Partitioning and replication” box to view when broker counts changed.

    image

Alerting

There are many types of Control Center alerts and many ways to configure them. Use the Alerts management page to define triggers and actions, or click on individual resources to setup alerts from there.

image
  1. This example already has pre-configured triggers and actions. View the Alerts Triggers screen, and click Edit against each trigger to see configuration details.

    • The trigger Under Replicated Partitions happens when a broker reports non-zero under replicated partitions, and it causes an action Email Administrator.
    • The trigger Consumption Difference happens when consumption difference for the Elasticsearch connector consumer group is greater than 0, and it causes an action Email Administrator.
    image
  2. If you followed the steps in the failed broker section, view the Alert history to see that the trigger Under Replicated Partitions happened and caused an alert when you stopped broker 2.

    image
  3. You can also trigger the Consumption Difference trigger. In the Kafka Connect -> Sinks screen, edit the running Elasticsearch sink connector.

  4. In the Connect view, pause the Elasticsearch sink connector in Settings by pressing the pause icon in the top right. This stops consumption for the related consumer group.

    image
  5. View the Alert history to see that this trigger happened and caused an alert.

    image

Monitoring

This tutorial has demonstrated how Confluent Control Center helps users manage their Confluent Platform deployment and how it provides monitoring capabilities for the cluster and applications. For most Confluent Platform users the Confluent Control Center monitoring and integrations are sufficent for production usage; however, some users wish to integrate with other monitoring solutions like Prometheus, Grafana, Datadog, and Splunk. The following JMX-based monitoring stacks help users setup a ‘single pane of glass’ monitoring solution for all their organization’s services and applications, including Kafka.

Here are some examples of monitoring stacks that integrate with Confluent Platform:

  1. JMX Exporter + Prometheus + Grafana (runnable with cp-demo from https://github.com/confluentinc/jmx-monitoring-stacks):

    image
    image
    image
  2. Jolokia + Elasticsearch + Kibana (runnable with cp-demo from https://github.com/confluentinc/jmx-monitoring-stacks):

    image
    image
    image
  3. Monitoring Confluent Platform with Datadog:

    image
  4. Next step: for a practical guide to optimizing your Kafka deployment for various service goals including throughput, latency, durability and availability, and useful metrics to monitor for performance and cluster health for on-prem Kafka clusters, see the Optimizing Your Apache Kafka Deployment whitepaper.

Troubleshooting

Here are some suggestions on how to troubleshoot the example.

  1. Verify the status of the Docker containers show Up state, except for the kafka-client container which is expected to have Exit 0 state.

    docker-compose ps
    

    Your output should resemble:

               Name                          Command                  State                                           Ports
    ------------------------------------------------------------------------------------------------------------------------------------------------------------
    connect                       bash -c sleep 10 && cp /us ...   Up             0.0.0.0:8083->8083/tcp, 9092/tcp
    control-center                /etc/confluent/docker/run        Up (healthy)   0.0.0.0:9021->9021/tcp, 0.0.0.0:9022->9022/tcp
    elasticsearch                 /bin/bash bin/es-docker          Up             0.0.0.0:9200->9200/tcp, 0.0.0.0:9300->9300/tcp
    kafka-client                  bash -c -a echo Waiting fo ...   Exit 0
    kafka1                        bash -c if [ ! -f /etc/kaf ...   Up (healthy)   0.0.0.0:10091->10091/tcp, 0.0.0.0:11091->11091/tcp, 0.0.0.0:12091->12091/tcp,
                                                                                  0.0.0.0:8091->8091/tcp, 0.0.0.0:9091->9091/tcp, 9092/tcp
    kafka2                        bash -c if [ ! -f /etc/kaf ...   Up (healthy)   0.0.0.0:10092->10092/tcp, 0.0.0.0:11092->11092/tcp, 0.0.0.0:12092->12092/tcp,
                                                                                  0.0.0.0:8092->8092/tcp, 0.0.0.0:9092->9092/tcp
    kibana                        /bin/sh -c /usr/local/bin/ ...   Up             0.0.0.0:5601->5601/tcp
    ksqldb-cli                    /bin/sh                          Up
    ksqldb-server                 /etc/confluent/docker/run        Up (healthy)   0.0.0.0:8088->8088/tcp
    openldap                      /container/tool/run --copy ...   Up             0.0.0.0:389->389/tcp, 636/tcp
    restproxy                     /etc/confluent/docker/run        Up             8082/tcp, 0.0.0.0:8086->8086/tcp
    schemaregistry                /etc/confluent/docker/run        Up             8081/tcp, 0.0.0.0:8085->8085/tcp
    streams-demo                  /app/start.sh                    Up             9092/tcp
    tools                         /bin/bash                        Up
    zookeeper                     /etc/confluent/docker/run        Up (healthy)   0.0.0.0:2181->2181/tcp, 2888/tcp, 3888/tcp
    
  2. If any containers are not in Up state, verify in the advanced Docker preferences settings that the memory available to Docker is at least 8 GB (default is 2 GB).

  3. For those Docker containers that are not in Up state, view the container’s logs with the command docker-compose logs [container] and look for error messages and exceptions.

    docker-compose logs
    
  4. To view sample messages for each topic, including wikipedia.parsed:

    ./scripts/consumers/listen.sh
    
  5. If a command that communicates with ZooKeeper appears to be failing with the error org.apache.zookeeper.KeeperException$NoAuthException, change the container you are running the command from to be either kafka1 or kafka2. This is because ZooKeeper is configured for SASL/DIGEST-MD5, and any commands that communicate with ZooKeeper need properties set for ZooKeeper authentication.

  6. Run any of the validation scripts to check that things are working.

    cd scripts/validate/
    
  7. If you are running cp-demo for a long time and issuing Confluent CLI commands on the tools container that return:

    Your token has expired. You are now logged out.
    Error: You must log in to run that command.
    

    Then run the following command, which refreshes the token by logging back in to MDS:

    ./scripts/helper/refresh_mds_login.sh
    

Teardown

  1. Stop the consumer group app to stop consuming from topic wikipedia.parsed. Note that the command below stops the consumers gracefully with kill -15, so the consumers follow the shutdown sequence.

    ./scripts/app/stop_consumer_app_group_graceful.sh
    
  2. Stop the Docker environment, destroy all components and clear all Docker volumes.

    ./scripts/stop.sh