Now, take the last step. In the previous recipes, it was explained how to produce messages from the console; this recipe indicates how to read the messages generated. Kafka also has a fancy command-line utility that enables consuming messages. Recall that all the command-line tasks can also be done programmatically. Also, recall that each line of the input was considered a message from the producer.
For this recipe, the execution of the previous recipes in this chapter is needed: Kafka already downloaded and installed, the Kafka nodes up and running, and a topic created inside the cluster. Also, some messages need to be produced with the message console producer. To begin consuming some messages from the console, change to the Kafka directory in the command line.
Consuming messages through the console is easy; just run the following command:
> ./bin/kafka-console-consumer.sh --topic humbleTopic --bootstrap-server localhost:9093 --from-beginning Her first word was Mom Her second word was Dad
The parameters are the topic and broker names of the producer. Also, the --from-begining parameter specifies that messages should be consumed from the beginning instead of the last messages in the log (go and give it a try: generate many more messages and don't specify this parameter).
There are more useful parameters for this command; some interesting ones are:
--fetch-size: The amount of data to be fetched in a single request. The size in bytes follows this argument. The default value is 1024 * 1024.--socket-buffer-size: The size of theTCP RECV. The size in bytes follows this argument. The default value is 2 * 1024 * 1024.--formater: The name of a class to use for formatting messages for display. The default value isNewlineMessageFormatter(already presented in the recipe).--autocommit.interval.ms: The time interval at which to save the current offset (the offset concept will be explained later) in milliseconds. The time in milliseconds follows the argument. The default value is 10,000.--max-messages: The maximum number of messages to consume before exiting. If not set, the consumption is continual. The number of messages follows the argument.--skip-message-on-error: If there is an error while processing a message, the system should skip it instead of halt.
Enough boring theory; this is a practical cookbook, so look at these most solicited menu entries:
Consume just one message:
> ./bin/kafka-console-consumer.sh --topic humbleTopic --bootstrap-server localhost:9093 --max-messages 1Consume one message from an offset:
> ./bin/kafka-console-consumer.sh --topic humbleTopic --bootstrap-server localhost:9093 --max-messages 1 --formatter 'kafka.coordinator.GroupMetadataManager$OffsetsMessageFormatter'Consume messages from a specific consumer group (consumer groups will be explained further):
> ./bin/kafka-console-consumer.sh --topic humbleTopic --bootstrap-server localhost:9093 --new-consumer --consumer-property group.id=my-groupIf you want to see the complete list of arguments, take a look at the command source code at: https://github.com/kafka-dev/kafka/blob/master/core/src/main/scala/kafka/consumer/ConsoleConsumer.scala