Updated on 2024-10-09 GMT+08:00

Connecting ClickHouse to Kafka in Normal Mode

This topic is available for MRS 3.3.0-LTS and later versions only.

Scenario

This topic describes how to connect to Kafka in normal mode and consume Kafka data.

Prerequisites

  • A Kafka cluster has been created and is in normal mode (Kerberos authentication is disabled).
  • You have created a ClickHouse cluster and installed the ClickHouse client. The ClickHouse and Kafka clusters can communicate with each other.

Procedure

  1. Log in to the FusionInsight Manager of the cluster where ClickHouse is deployed, and choose Cluster > Services > ClickHouse. Click Configurations then All Configurations, and click ClickHouseServer(Role) > Engine. The following table shows the parameter needs to be configured.

    Parameter

    Description

    kafka_auth_mode

    Authentication method for the connection between ClickHouse and Kafka. Set this parameter to NoAuth.

  2. Choose Cluster > Services > ClickHouse. Click Configurations > All Configurations > ClickHouseServer (Role) > Customization, and add the following parameters to clickhouse-config-customize.

    Name

    Value

    kafka.security_protocol

    plaintext

  3. Click Save. In the displayed dialog box, click OK to save the configuration. Choose Instances, select ClickHouseServer, and click More > Instance Rolling Restart.
  4. Go to the Kafka client installation directory. For details, see Using the Kafka Client.

    1. Log in to the node where the Kafka client is installed as the Kafka client installation user.
    2. Run the following command to go to the client installation directory:

      cd /opt/client

    3. Run the following command to configure environment variables:

      source bigdata_env

    4. If Kerberos authentication is enabled for the cluster, run the following command to authenticate the user. If Kerberos authentication is disabled for the cluster, skip this step.

      kinit Component service user

  5. Run the following command to create a Kafka topic. For details, see Creating a Kafka Topic.

    kafka-topics.sh --topic topic1 --create --zookeeper IP address of the Zookeeper role instance:Port used by ZooKeeper to listen to client/kafka --partitions 2 --replication-factor 1

    • --topic: name of the topic to be created, for example, topic1.
    • --zookeeper is the IP address of the node where the ZooKeeper role instances are located, which can be the IP address of any of the three role instances. You can obtain the IP address of the node by performing the following steps:

      Log in to FusionInsight Manager, choose Cluster > Services > ZooKeeper. On the page that is displayed, click the Instances tab to query the IP address of the ZooKeeper instance.

    • --partitions and --replication-factor are the topic partitions and topic backup replicas, respectively. The number of the two parameters cannot exceed the number of Kafka role instances.
    • To obtain the Port used by ZooKeeper to listen to the client, log in to FusionInsight Manager, click Cluster, choose Services > ZooKeeper, and view the value of clientPort on the Configuration tab page. The default port is 24002.

  6. Log in to the ClickHouse client node and connect it to the ClickHouse server. For details, see ClickHouse Client Practices.
  7. Create a Kafka table engine. The following is an example:

    CREATE TABLE queue1 (
    key String,
    value String,
    event_date DateTime
    ) ENGINE = Kafka()
    SETTINGS kafka_broker_list = 'kafka_ip1:21005,kafka_ip2:21005,kafka_ip3:21005',
    kafka_topic_list = 'topic1',
    kafka_group_name = 'group2',
    kafka_format = 'CSV',
    kafka_row_delimiter = '\n',
    kafka_handle_error_mode='stream';

    The following table lists the related parameters.

    Parameter

    Description

    kafka_broker_list

    A list of IP addresses and port numbers of Kafka broker instances. For example, :IP address 1 of Kafka broker instance:9092,IP address 2 of Kafka broker instance:9092,IP address 3 of Kafka broker instance:9092

    To obtain the IP address of the Kafka broker instance, perform the following steps:

    Log in to FusionInsight Manager and choose Cluster > Services > Kafka. Click the Instance tab to query the IP addresses of the Kafka instances.

    kafka_topic_list

    Topic where Kafka data is consumed

    kafka_group_name

    Kafka consumer group

    kafka_format

    Formatting type of consumed data. JSONEachRow indicates the JSON format (a piece of data in each line). CSV indicates the data is in a line but separated by commas (,). For more information, visit https://clickhouse.tech/docs/en/interfaces/formats/.

    kafka_row_delimiter

    Delimiter character, which ends a message.

    kafka_handle_error_mode

    If this parameter is set to stream, each message processing exception is printed. You need to create a view and query the specific exception of abnormal data through the view.

    The following example shows you how to create a view:

    CREATE MATERIALIZED VIEW default.kafka_errors2
    (
    `topic` String,
    `key` String,
    `partition` Int64,
    `offset` Int64,
    `timestamp` Date,
    `timestamp_ms` Int64,
    `raw` String,
    `error` String
    )
    ENGINE = MergeTree
    ORDER BY (topic, partition, offset)
    SETTINGS index_granularity = 8192 AS
    SELECT
    _topic AS topic,
    _key  AS key,
    _partition AS partition,
    _offset AS offset,
    _timestamp AS timestamp,
    _timestamp_ms AS timestamp_ms,
    _raw_message AS raw,
    _error AS error
    FROM default.queue1;

    Query the view. The following is an example:

    host1 :) select * from kafka_errors2;  SELECT * FROM kafka_errors2  Query id: bf4d788f-bcb9-44f5-95d0-a6c83c591ddb  ┌─topic──┬─key─┬─partition─┬─offset─┬──timestamp─┬─timestamp_ms─┬─raw─┬─error────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐ │ topic1 │     │         1 │      8 │ 2023-06-20 │   1687252213 │ 456 │ Cannot parse date: value is too short: (at row 1) Buffer has gone, cannot extract information about what has been parsed. │ └────────┴─────┴───────────┴────────┴────────────┴──────────────┴─────┴──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘  1 rows in set. Elapsed: 0.003 sec.   host1 :)

    kafka_skip_broken_messages

    (Optional) Number of Kafka data records where parsing exceptions are ignored. If N exceptions occur and the background thread ends, the materialized view is re-arranged to monitor the data.

    kafka_num_consumers

    (Optional) Number of consumers of a single Kafka engine. You can set this parameter to a larger value to improve the data consumption throughput. But the maximum value of this parameter cannot exceed the total number of partitions of the corresponding topic.

    For details about other configurations, see https://clickhouse.com/docs/en/engines/table-engines/integrations/kafka.

  8. Connect the client to ClickHouse to create a local table. The following is an example:

    CREATE TABLE daily1(
    key String,
    value String,
    event_date DateTime
    )ENGINE = MergeTree()
    ORDER BY key;

  9. Connect the client to ClickHouse to create a materialized view. The following is an example:

    CREATE MATERIALIZED VIEW default.consumer1 TO default.daily1 (
    `event_date` DateTime,
    `key` String,
    `value` String
    ) AS
    SELECT
    event_date,
    key,
    value
    FROM default.queue1;

  10. Perform 4 again to go to the Kafka client installation directory.
  11. Run the following command to send a message to the topic created in 5:

    kafka-console-producer.sh --broker-list IP address 1 of the Kafka broker instance:9092,IP address 2 of the Kafka broker instance:9092,IP address 3 of the Kafka broker instance:9092 --topic topic1
    >a1,b1,'2020-08-01 10:00:00'
    >a2,b2,'2020-08-02 10:00:00'
    >a3,b3,'2020-08-02 10:00:00'
    >a4,b4,'2023-09-02 10:00:00'

  12. Query the consumed Kafka data and the preceding materialized view. The following is an example:

    select * from daily;