Updated on 2023-11-17 GMT+08:00

HBase Source Table

Function

Create a source stream to obtain data from HBase as input for jobs. HBase is a column-oriented distributed cloud storage system that features enhanced reliability, excellent performance, and elastic scalability. It applies to the storage of massive amounts of data and distributed computing. You can use HBase to build a storage system capable of storing TB- or even PB-level data. With HBase, you can filter and analyze data with ease and get responses in milliseconds, rapidly mining data value. DLI can read data from HBase for filtering, analysis, and data dumping.

Prerequisites

  • An enhanced datasource connection has been created for DLI to connect to HBase, so that jobs can run on the dedicated queue of DLI and you can set the security group rules as required.
  • If MRS HBase is used, IP addresses of all hosts in the MRS cluster have been added to host information of the enhanced datasource connection.

    .

    For details, see section "Modifying the Host Information" in the Data Lake Insight User Guide.

  • In Flink cross-source development scenarios, there is a risk of password leakage if datasource authentication information is directly configured. You are advised to use the datasource authentication provided by DLI.

    For details about datasource authentication, see Introduction to Datasource Authentication.

Precautions

  • When creating a Flink OpenSource SQL job, you need to set Flink Version to 1.12 on the Running Parameters tab of the job editing page, select Save Job Log, and set the OBS bucket for saving job logs.
  • The column families in created HBase source table must be declared as the ROW type, the field names map the column family names, and the nested field names map the column qualifier names.

    There is no need to declare all the families and qualifiers in the schema. Users can declare what is used in the query. Except the ROW type fields, the single atomic type field (for example, STRING or BIGINT) will be recognized as the HBase rowkey. The rowkey field can be an arbitrary name, but should be quoted using backticks if it is a reserved keyword.

Syntax

create table hbaseSource (
  attr_name attr_type 
  (',' attr_name attr_type)* 
  (',' watermark for rowtime_column_name as watermark-strategy_expression)
  ','PRIMARY KEY (attr_name, ...) NOT ENFORCED)
)
with (
  'connector' = 'hbase-2.2',
  'table-name' = '',
  'zookeeper.quorum' = ''
);

Parameters

Table 1 Parameter description

Parameter

Mandatory

Default Value

Data Type

Description

connector

Yes

None

String

Connector to be used. Set this parameter to hbase-2.2.

table-name

Yes

None

String

Name of the HBase table to connect.

zookeeper.quorum

Yes

None

String

HBase ZooKeeper quorum, in the format of "ZookeeperAddress:ZookeeperPort".

The following uses an MRS HBase cluster as an example to describe how to obtain the IP address and port number of ZooKeeper used by this parameter:

  • On MRS Manager, choose Cluster and click the name of the desired cluster. Choose Services > ZooKeeper > Instance, and obtain the IP address of the ZooKeeper instance.
  • On MRS Manager, choose Cluster and click the name of the desired cluster. Choose Services > ZooKeeper > Configurations > All Configurations, search for the clientPort parameter, and obtain its value, that is, the ZooKeeper port number.

zookeeper.znode.parent

No

/hbase

String

Root directory in ZooKeeper. The default value is /hbase.

null-string-literal

No

None

String

Representation for null values for string fields.

HBase source encodes/decodes empty bytes as null values for all types except the string type.

krb_auth_name

No

None

String

Name of datasource authentication of the Kerberos type created on DLI.

Data Type Mapping

HBase stores all data as byte arrays. The data needs to be serialized and deserialized during read and write operations.

When serializing and de-serializing, Flink HBase connector uses utility class org.apache.hadoop.hbase.util.Bytes provided by HBase (Hadoop) to convert Flink data types to and from byte arrays.

Flink HBase connector encodes null values to empty bytes, and decode empty bytes to null values for all data types except the string type. For the string type, the null literal is determined by the null-string-literal option.

Table 2 Data type mapping

Flink SQL Type

HBase Conversion

CHAR/VARCHAR/STRING

byte[] toBytes(String s)

String toString(byte[] b)

BOOLEAN

byte[] toBytes(boolean b)

boolean toBoolean(byte[] b)

BINARY/VARBINARY

Returns byte[] as is.

DECIMAL

byte[] toBytes(BigDecimal v)

BigDecimal toBigDecimal(byte[] b)

TINYINT

new byte[] { val }

bytes[0] // returns first and only byte from bytes

SMALLINT

byte[] toBytes(short val)

short toShort(byte[] bytes)

INT

byte[] toBytes(int val)

int toInt(byte[] bytes)

BIGINT

byte[] toBytes(long val)

long toLong(byte[] bytes)

FLOAT

byte[] toBytes(float val)

float toFloat(byte[] bytes)

DOUBLE

byte[] toBytes(double val)

double toDouble(byte[] bytes)

DATE

Stores the number of days since epoch as an int value.

TIME

Stores the number of milliseconds of the day as an int value.

TIMESTAMP

Stores the milliseconds since epoch as a long value.

ARRAY

Not supported

MAP/MULTISET

Not supported

ROW

Not supported

Example

In this example, data is read from the HBase data source and written to the Print result table. The procedure is as follows (the HBase versions used in this example are 1.3.1, 2.1.1, and 2.2.3):

  1. Create an enhanced datasource connection in the VPC and subnet where HBase locates, and bind the connection to the required Flink queue. For details, see Enhanced Datasource Connections. .
  2. Set HBase cluster security groups and add inbound rules to allow access from the Flink job queue. Test the connectivity using the HBase address by referring to Testing Address Connectivity. If the connection is successful, the datasource is bound to the queue. Otherwise, the binding fails.
  3. Use the HBase shell to create HBase table order that has only one column family detail. The creation statement is as follows:
    create 'order', {NAME => 'detail'}
  4. Run the following command in the HBase shell to insert a data record:
    put 'order', '202103241000000001', 'detail:order_channel','webShop'
    put 'order', '202103241000000001', 'detail:order_time','2021-03-24 10:00:00'
    put 'order', '202103241000000001', 'detail:pay_amount','100.00'
    put 'order', '202103241000000001', 'detail:real_pay','100.00'
    put 'order', '202103241000000001', 'detail:pay_time','2021-03-24 10:02:03'
    put 'order', '202103241000000001', 'detail:user_id','0001'
    put 'order', '202103241000000001', 'detail:user_name','Alice'
    put 'order', '202103241000000001', 'detail:area_id','330106'
  5. Create a Flink OpenSource SQL job. Enter the following job script and submit the job. The job script uses the HBase data source and the Print result table.
    When you create a job, set Flink Version to 1.12 on the Running Parameters tab. Select Save Job Log, and specify the OBS bucket for saving job logs. Change the values of the parameters in bold as needed in the following script.
    create table hbaseSource (
      order_id string,-- Indicates the unique rowkey.
      detail Row( -- Indicates the column family.
        order_channel string,
        order_time string,
        pay_amount string,
        real_pay string,
        pay_time string,
        user_id string,
        user_name string,
        area_id string),
      primary key (order_id) not enforced
    ) with (
      'connector' = 'hbase-2.2',
       'table-name' = 'order',
       'zookeeper.quorum' = 'ZookeeperAddress:ZookeeperPort'
    ) ;
    
    create table printSink (
      order_id string,
      order_channel string,
      order_time string,
      pay_amount string,
      real_pay string,
      pay_time string,
      user_id string,
      user_name string,
      area_id string
    ) with (
     'connector' = 'print'
    );
    
    insert into printSink select order_id, detail.order_channel,detail.order_time,detail.pay_amount,detail.real_pay,
    detail.pay_time,detail.user_id,detail.user_name,detail.area_id from hbaseSource;
  6. Perform the following operations to view the data result in the taskmanager.out file:
    1. Log in to the DLI console. In the navigation pane, choose Job Management > Flink Jobs.
    2. Click the name of the corresponding Flink job, choose Run Log, click OBS Bucket, and locate the folder of the log you want to view according to the date.
    3. Go to the folder of the date, find the folder whose name contains taskmanager, download the taskmanager.out file, and view result logs.

    The data result is as follows:

    +I(202103241000000001,webShop,2021-03-24 10:00:00,100.00,100.00,2021-03-24 10:02:03,0001,Alice,330106)

FAQ

  • Q: What should I do if the Flink job execution fails and the log contains the following error information?
    java.lang.IllegalArgumentException: offset (0) + length (8) exceed the capacity of the array: 6

    A: If data in the HBase table is imported in other modes, the data is represented in the string format. Therefore, this error is reported when other data formats are used. Change the type of the non-string fields in the HBase source table created by Flink to the string format.

  • Q: What should I do if the Flink job execution fails and the log contains the following error information?
    org.apache.zookeeper.ClientCnxn$SessionTimeoutException: Client session timed out, have not heard from server in 90069ms for connection id 0x0

    A: The datasource connection is not bound, the binding fails, or the security group of the HBase cluster is not configured to allow access from the network segment of the DLI queue. Configure the datasource connection by referring to Enhanced Datasource Connection or configure the security group of the HBase cluster to allow access from the DLI queue.