两只松鼠的故事：flink-connector-opengauss

Posted 2021-10-20 麒思妙想

tags:

篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了两只松鼠的故事：flink-connector-opengauss相关的知识，希望对你有一定的参考价值。

感谢松鼠会大佬的再三邀请。对我来说这算是一篇命题作文，那么我的答案是什么呢？刚好我也很喜欢另外一个松鼠社区，那么就用两只松鼠来做答案吧，没错，Flink和OpenGauss就是我的答案：

手把手完成一次OpenGauss的安装
手把手构建 Flink connector opengauss

OpenGauss 安装

操作系统建议使用 openEuler20.03LTS ，注意不要使用sp包，可能会因为sp包依赖升级，而导致安装失败。使用下面安装脚本，我对官方脚本进行了一些修改，适合做单机实验，不再需要手动设置ip，而是直接将 eth0 的 ip 作为数据库ip使用。

#!/bin/bash

## Author：  贾军锋
## update：  dafei1288 @ 2021-09-27
## Date：    2021-04-15
## OS:       openEuler20.03LTS [最小硬件配置：2c/4G]
## Database：openGauss 2.0.0
## Description：一键式实现操作系统环境配置、openGauss软件下载、openGauss软件安装等步骤，帮助大家提升安装openGauss数据库效率
## Tips:     请确保操作系统可以连接外网

## 0.关闭virbr0网卡 [本地虚拟机标准化安装openEuler系统会默认存在virbr0网卡，删除该网卡以避免干扰数据库的安装]
## virsh net-destroy default
## virsh net-list
## echo "Net device virbr0 is disabled."


## 1.定义主机信息[请根据实际情况修改]
export MY_HOSTNAME=node1           ## 主机名
export MY_HOSTIP=$(ifconfig eth0 | grep 'inet' | awk '{print $2}' | head -1)     ## IP地址
export MY_SOFTWARE_DIRECTORY=/soft/openGauss      ## 软件包所在目录
export MY_XML=/soft/openGauss/clusterconfig.xml   ## 集群配置文件XML
export openGauss_Download_url=https://opengauss.obs.cn-south-1.myhuaweicloud.com/2.0.0/x86_openEuler/openGauss-2.0.0-openEuler-64bit-all.tar.gz  ## openGauss软件包下载地址

## 1. 设置主机名并配置hosts文件
hostnamectl set-hostname $MY_HOSTNAME
sed -i '/$MY_HOSTIP/d' /etc/hosts
echo "$MY_HOSTIP  $MY_HOSTNAME   #Gauss OM IP Hosts Mapping" >> /etc/hosts
cat /etc/hosts
echo "1.Configure /etc/hosts completed."
echo -e "\\n"

## 2. 关闭防火墙
systemctl disable firewalld.service
systemctl stop firewalld.service
echo "Firewalld " `systemctl status firewalld|grep Active`
echo "2.Disable firewalld service completed."
echo -e "\\n"

## 3. 关闭SELinux
sed -i '/^SELINUX=/d' /etc/selinux/config
echo "SELINUX=disabled" >> /etc/selinux/config
cat /etc/selinux/config|grep "SELINUX=disabled"
echo "3.Disable SELINUX completed."
echo -e "\\n"


## 4. 设置操作系统字符集编码
echo "LANG=en_US.UTF-8" >> /etc/profile
source /etc/profile
echo $LANG
echo "4.Configure encoding completed."
echo -e "\\n"

## 5. 设置操作系统时区
rm -fr /etc/localtime
ln -s /usr/share/zoneinfo/Asia/Shanghai  /etc/localtime
date -R
hwclock
echo "5.Configure Timezone completed."
echo -e "\\n"

## 6. 关闭SWAP分区 [对于2G内存的设备，建议待安装完毕后再打开SWAP以间接 “扩容内存容量”]
sed -i '/swap/s/^/#/' /etc/fstab
swapoff -a
free -m
echo "6.Close swap partition completed."
echo -e "\\n"


## 7. 配置SSH服务，关闭Banner，允许root远程登录
sed -i '/Banner/s/^/#/'  /etc/ssh/sshd_config
sed -i '/PermitRootLogin/s/^/#/'  /etc/ssh/sshd_config
echo -e "\\n" >> /etc/ssh/sshd_config
echo "Banner none " >> /etc/ssh/sshd_config
echo "PermitRootLogin yes" >> /etc/ssh/sshd_config
cat /etc/ssh/sshd_config |grep -v ^#|grep -E 'PermitRoot|Banner'
echo "7.Configure SSH Service completed."
echo -e "\\n"

## 8. 配置YUM源、安装依赖包、修改默认Python3版本
mkdir /etc/yum.repos.d/bak
mv /etc/yum.repos.d/*.repo  /etc/yum.repos.d/bak/
wget -O /etc/yum.repos.d/openEulerOS.repo https://repo.huaweicloud.com/repository/conf/openeuler_x86_64.repo
yum clean all
yum install -y bzip2 python3
yum install -y libaio-devel libnsl flex bison ncurses-devel glibc-devel patch readline-devel net-tools tar
mv /usr/bin/python  /usr/bin/python2_bak
ln -s /usr/bin/python3 /usr/bin/python
python -V
echo "8.Configure Install Packages and change default Python version completed."
echo -e "\\n"


## 9. 配置 sysctl.conf 和 performance.sh
cat >> /etc/sysctl.conf << EOF
net.ipv4.tcp_retries1 = 5
net.ipv4.tcp_syn_retries = 5
net.sctp.path_max_retrans = 10
net.sctp.max_init_retransmits = 10
EOF
sysctl -p

sed -i '/vm.min_free_kbytes/s/^/#/' /etc/profile.d/performance.sh   ## Only for openEuler
cat /etc/profile.d/performance.sh|grep vm.min_free_kbytes

echo "9.Configure sysctl.conf and performance.sh completed."
echo -e "\\n"


## 10. 配置资源限制
echo "* soft stack 3072" >> /etc/security/limits.conf
echo "* hard stack 3072" >> /etc/security/limits.conf
echo "* soft nofile 1000000" >> /etc/security/limits.conf
echo "* hard nofile 1000000" >> /etc/security/limits.conf
echo "* soft nproc unlimited" >> /etc/security/limits.d/90-nproc.conf
tail -n 4 /etc/security/limits.conf
tail -n 1 /etc/security/limits.d/90-nproc.conf
echo "10.Configure resource limits completed."
echo -e "\\n"

## 11. 关闭透明大页[Only for CentOS]
cat >>/etc/rc.d/rc.local<<EOF
if test -f /sys/kernel/mm/transparent_hugepage/enabled; then
   echo never > /sys/kernel/mm/transparent_hugepage/enabled
fi
if test -f /sys/kernel/mm/transparent_hugepage/defrag; then
   echo never > /sys/kernel/mm/transparent_hugepage/defrag
fi
EOF
chmod +x /etc/rc.d/rc.local
/usr/bin/sh /etc/rc.d/rc.local
cat /sys/kernel/mm/transparent_hugepage/enabled
cat /sys/kernel/mm/transparent_hugepage/defrag
echo "11.Close transparent_hugepage completed."
echo -e "\\n"


## 12. 禁用RemoveIPC[Only for openEuler]
sed -i '/^RemoveIPC/d' /etc/systemd/logind.conf
sed -i '/^RemoveIPC/d' /usr/lib/systemd/system/systemd-logind.service
echo "RemoveIPC=no"  >> /etc/systemd/logind.conf
echo "RemoveIPC=no"  >> /usr/lib/systemd/system/systemd-logind.service
systemctl daemon-reload
systemctl restart systemd-logind
loginctl show-session | grep RemoveIPC
systemctl show systemd-logind | grep RemoveIPC
echo "12.Disable RemoveIPC completed."
echo -e "\\n"


## 13. 下载openGauss软件包
mkdir -p $MY_SOFTWARE_DIRECTORY
cd $MY_SOFTWARE_DIRECTORY
wget $openGauss_Download_url
echo "13.openGauss software download completed."
echo -e "\\n"

## 14. 配置XML文件
rm -fr $MY_XML
cat >> $MY_XML <<EOF
<?xml version="1.0" encoding="UTF-8"?> 
<ROOT> 
    <!-- openGauss整体信息 --> 
    <CLUSTER> 
        <PARAM name="clusterName" value="dbCluster" /> 
        <PARAM name="nodeNames" value="$MY_HOSTNAME" /> 
        <PARAM name="backIp1s" value="$MY_HOSTIP"/> 
        <PARAM name="gaussdbAppPath" value="/gaussdb/app" /> 
        <PARAM name="gaussdbLogPath" value="/gaussdb/log" /> 
        <PARAM name="gaussdbToolPath" value="/gaussdb/om" /> 
        <PARAM name="corePath" value="/gaussdb/corefile"/> 
        <PARAM name="clusterType" value="single-inst"/> 
    </CLUSTER> 
    <!-- 每台服务器上的节点部署信息 --> 
    <DEVICELIST> 
        <!-- node1上的节点部署信息 --> 
        <DEVICE sn="1000001"> 
            <PARAM name="name" value="$MY_HOSTNAME"/> 
            <PARAM name="azName" value="AZ1"/> 
            <PARAM name="azPriority" value="1"/> 
            <!-- 如果服务器只有一个网卡可用，将backIP1和sshIP1配置成同一个IP --> 
            <PARAM name="backIp1" value="$MY_HOSTIP"/> 
            <PARAM name="sshIp1" value="$MY_HOSTIP"/> 
             
     <!--dbnode--> 
     <PARAM name="dataNum" value="1"/> 
     <PARAM name="dataPortBase" value="26000"/> 
     <PARAM name="dataNode1" value="/gaussdb/data/db1"/> 
        </DEVICE> 
    </DEVICELIST> 
</ROOT>
EOF
cat $MY_XML
echo "14.Configure XML file completed."
echo -e "\\n"


## 15. 解压安装包并修改目录权限
echo "Begin to Uncompress openGauss Package and Modify directory permissions:"
cd $MY_SOFTWARE_DIRECTORY
tar -zxvf *all.tar.gz
tar -zxvf *om.tar.gz
ls -l
chmod -R 777 $MY_SOFTWARE_DIRECTORY/../
echo "15.Uncompress openGauss Package completed."
echo -e "\\n"

## 16. 执行 gs_preinstall
echo "Begin to execute openGauss preinstall:"
python $MY_SOFTWARE_DIRECTORY/script/gs_preinstall -U omm -G dbgrp -X $MY_XML
echo "16.openGauss preinstall completed."
echo -e "\\n"


## 17. 检查预安装环境
echo "Begin to Check OS environment:"
$MY_SOFTWARE_DIRECTORY/script/gs_checkos -i A -h $MY_HOSTNAME --detail

## 18. 执行 gs_install
echo "Begin to execute openGauss install:"
touch /home/omm/install_db
cat >> /home/omm/install_db <<EOF
source ~/.bashrc
gs_install -X  $MY_XML --gsinit-parameter="--encoding=UTF8"  --dn-guc="max_process_memory=2GB" --dn-guc="shared_buffers=128MB" --dn-guc="cstore_buffers=16MB"
EOF
chown -R omm:dbgrp /home/omm/install_db
su - omm -c "sh /home/omm/install_db"
echo "17.openGauss install completed."
echo -e "\\n"

## 安装完毕！
echo "openGauss Install completed.congratulations"
echo "Congratulations!!!"

另外，如果你也是虚拟机安装，注意至少分配2核，4G内存（上图是我的虚拟机配置，仅供参考），否则会出现如下错误：

[GAUSS-51400] : Failed to execute the command: source /home/omm/.bashrc;python3 '/gaussdb/om/script/local/InitInstance.py' -U omm  -P "--encoding=UTF8" -l /gaussdb/log/omm/om/gs_local.log.Error:
Initializing instance.
[GAUSS-51615] : Failed to initialize instance. Command:/gaussdb/app/bin/gs_initdb --locale=C -D /gaussdb/data/db1 --nodename=dn_6001 --encoding=UTF8 -C /gaussdb/app/bin. Error:
The files belonging to this database system will be owned by user "omm".
This user must also own the server

OpenGauss 常用命令

切换至数据库用户

su - omm

查看数据库状态

gs_om -t status --all

[omm@node1 ~]$ gs_om -t status --all
-----------------------------------------------------------------------

cluster_state             : Normal
redistributing            : No

-----------------------------------------------------------------------

node                      : 1
node_name                 : node1
instance_id               : 6001
node_ip                   : 172.21.235.7
data_path                 : /gaussdb/data/db1
type                      : Datanode
instance_state            : Normal
az_name                   : AZ1
instance_role             : Normal

-----------------------------------------------------------------------

重启服务命令

gs_om -t restart

[omm@node1 ~]$  gs_om -t restart
Stopping cluster.
=========================================
Successfully stopped cluster.
=========================================
End stop cluster.
Starting cluster.
=========================================
[SUCCESS] node1
2021-09-27 15:53:05.285 61517861.1 [unknown] 140634486265600 [unknown] 0 dn_6001 01000  0 [BACKEND] WARNING:  could not create any HA TCP/IP sockets
2021-09-27 15:53:05.300 61517861.1 [unknown] 140634486265600 [unknown] 0 dn_6001 01000  0 [BACKEND] WARNING:  Failed to initialize the memory protect for g_instance.attr.attr_storage.cstore_buffers (16 Mbytes) or shared memory (1496 Mbytes) is larger.
=========================================
Successfully started.

连接数据库

gsql -d postgress -p 26000 -r

[omm@node1 ~]$ gsql -d postgres -p 26000 -r
gsql ((openGauss 2.0.0 build 78689da9) compiled at 2021-03-31 21:04:06 commit 0 last mr  )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.

postgres=#

创建用户

postgres=# create user jacky with password '123456';
CREATE ROLE

创建数据库

postgres=# CREATE DATABASE test OWNER jacky;

授权

postgres=# GRANT ALL PRIVILEGES ON DATABASE test TO jacky;

退出命令行

postgres=# \\q

配置 navicat 连接 OpenGauss

这里配置稍微有一点繁琐，需要开启用户的md5验证，然后才能使用navicat进行管理。开启md5验证需要如下步骤：

首先：vi /gaussdb/data/db1/pg_hba.conf

# TYPE  DATABASE        USER            ADDRESS                 METHOD

# "local" is for Unix domain socket connections only
local   all             all                                     trust
# IPv4 local connections:
host    all             all             127.0.0.1/32            trust
host    all    all    172.27.71.161/32    trust
host    all     all     172.27.64.1/32          md5
# IPv6 local connections:
host    all             all             ::1/128                 trust
# Allow replication connections from localhost, by a user with the
# replication privilege.
#local   replication     omm                                trust
#host    replication     omm        127.0.0.1/32            trust
#host    replication     omm        ::1/128                 trust

添加一行数据，其中 172.27.64.1 替换为，你的客户端IP

host all all 172.27.64.1/32 md5

然后 vi /gaussdb/data/db1/postgresql.conf

modify_initial_password = true  #Whether to change the initial password of the initial user
#password_policy = 1                    #Whether password complexity checks
#password_reuse_time = 60               #Whether the new password can be reused in password_reuse_time days
#password_reuse_max = 0                 #Whether the new password can be reused
#password_lock_time = 1                 #The account will be unlocked automatically after a specified period of time
#failed_login_attempts = 10             #Enter the wrong password reached failed_login_attempts times, the current account will be locked
password_encryption_type = 0            #Password storage type, 0 is md5 for PG, 1 is sha256 + md5, 2 is sha256 only
#password_min_length = 8                #The minimal password length(6-999)
#password_max_length = 32               #The maximal password length(6-999)
#password_min_uppercase = 0             #The minimal upper character number in password(0-999)
#password_min_lowercase = 0             #The minimal lower character number in password(0-999)
#password_min_digital = 0               #The minimal digital character number in password(0-999)
#password_min_special = 0               #The minimal special character number in password(0-999)
#password_effect_time = 90d             #The password effect time(0-999)
#password_notify_time = 7d              #The password notify time(0-999)

找到 password_encryption_type 修改为 0

重启服务即可

Flink connector opengauss

我们其实是可以自己手写Sink将CDC的数据直接汇入我们的目标数据库的。这样是不是不够优雅？我们是不是可以通过Flink SQL的方式将数据汇入到opengauss呢？答案是肯定的，接下来我们就来实现一个简单的opengauss的Flink connector

构建行转换器（RowConverter）
构建方言（Dialect）
注册动态表工厂（DynamicTableFactory），以及相关Sink程序

经过上面三步，就可以实现一个简单的connector了。接下来我们就来看，如何实现:

构建行转换器（RowConverter）

package name.lijiaqi.converter;

import org.apache.flink.connector.jdbc.internal.converter.AbstractJdbcRowConverter;
import org.apache.flink.table.types.logical.RowType;

/**
 * @author lijiaqi
 */
public class OpenGaussRowConverter extends AbstractJdbcRowConverter {

    public OpenGaussRowConverter(RowType rowType) {
        super(rowType);
    }

    private static final long serialVersionUID = 1L;

    @Override
    public String converterName() {
        return "opengauss";
    }

}

构建方言（Dialect）

package name.lijiaqi.dialect;

import name.lijiaqi.converter.OpenGaussRowConverter;
import org.apache.flink.connector.jdbc.dialect.JdbcDialect;
import org.apache.flink.connector.jdbc.internal.converter.JdbcRowConverter;
import org.apache.flink.table.api.TableSchema;
import org.apache.flink.table.api.ValidationException;
import org.apache.flink.table.types.logical.RowType;

import java.util.Optional;

/**
 *
 * @author lijiaqi
 */
public class OpenGaussDialect implements JdbcDialect {

    private static final long serialVersionUID = 1L;

    @Override
    public String dialectName() {
        return "opengauss";
    }

    @Override
    public boolean canHandle(String url) {
        return url.startsWith("jdbc:opengauss:");
    }

    @Override
    public JdbcRowConverter getRowConverter(RowType rowType) {
        return new OpenGaussRowConverter(rowType);
    }

    @Override
    public String getLimitClause(long l) {
        return null;
    }

    @Override
    public void validate(TableSchema schema) throws ValidationException {
        JdbcDialect.super.validate(schema);
    }

    @Override
    public Optional<String> defaultDriverName() {
        return Optional.of("org.opengauss.Driver");
    }

    @Override
    public String quoteIdentifier(String identifier) {
        return "'" + identifier + "'";
    }

    @Override
    public Optional<String> getUpsertStatement(String tableName, String[] fieldNames, String[] uniqueKeyFields) {
        return JdbcDialect.super.getUpsertStatement(tableName, fieldNames, uniqueKeyFields);
    }

    @Override
    public String getRowExistsStatement(String tableName, String[] conditionFields) {
        return JdbcDialect.super.getRowExistsStatement(tableName, conditionFields);
    }

    @Override
    public String getInsertIntoStatement(String tableName, String[] fieldNames) {
        return JdbcDialect.super.getInsertIntoStatement(tableName, fieldNames);
    }

    @Override
    public String getUpdateStatement(String tableName, String[] fieldNames, String[] conditionFields) {
        return JdbcDialect.super.getUpdateStatement(tableName, fieldNames, conditionFields);
    }

    @Override
    public String getDeleteStatement(String tableName, String[] conditionFields) {
        return JdbcDialect.super.getDeleteStatement(tableName, conditionFields);
    }

    @Override
    public String getSelectFromStatement(String tableName, String[] selectFields, String[] conditionFields) {
        return JdbcDialect.super.getSelectFromStatement(tableName, selectFields, conditionFields);
    }

}

注册动态表工厂（DynamicTableFactory），以及相关Sink程序

首先创建 OpenGaussSinkFunction 用于接受RowData数据输入，并将其Sink到配置的数据库中

package name.lijiaqi.table;

import org.apache.flink.api.common.serialization.SerializationSchema;
import org.apache.flink.connector.jdbc.internal.options.JdbcOptions;
import org.apache.flink.table.connector.ChangelogMode;
import org.apache.flink.table.connector.format.EncodingFormat;
import org.apache.flink.table.connector.sink.DynamicTableSink;
import org.apache.flink.table.connector.sink.SinkFunctionProvider;
import org.apache.flink.table.data.RowData;
import org.apache.flink.table.types.DataType;

/**
 * @author lijiaqi
 */
public class OpenGaussDynamicTableSink implements DynamicTableSink {

    private final JdbcOptions jdbcOptions;
    private final EncodingFormat<SerializationSchema<RowData>> encodingFormat;
    private final DataType dataType;

    public OpenGaussDynamicTableSink(JdbcOptions jdbcOptions, EncodingFormat<SerializationSchema<RowData>> encodingFormat, DataType dataType) {
        this.jdbcOptions = jdbcOptions;
        this.encodingFormat = encodingFormat;
        this.dataType = dataType;
    }

    @Override
    public ChangelogMode getChangelogMode(ChangelogMode requestedMode) {
        return requestedMode;
    }

    @Override
    public SinkRuntimeProvider getSinkRuntimeProvider(Context context) {
        System.out.println("SinkRuntimeProvider");
        System.out.println(dataType);

//        SerializationSchema<RowData> serializationSchema = encodingFormat.createRuntimeEncoder(context, dataType);
        OpenGaussSinkFunction gbasedbtSinkFunction = new OpenGaussSinkFunction(jdbcOptions,dataType);
        return SinkFunctionProvider.of(gbasedbtSinkFunction);
    }

    @Override
    public DynamicTableSink copy() {
        return new OpenGaussDynamicTableSink(jdbcOptions, encodingFormat, dataType);
    }

    @Override
    public String asSummaryString() {
        return "OpenGauss Table Sink";
    }

}

构建 OpenGaussDynamicTableSink

package name.lijiaqi.table;

import org.apache.flink.api.common.serialization.SerializationSchema;
import org.apache.flink.connector.jdbc.internal.options.JdbcOptions;
import org.apache.flink.table.connector.ChangelogMode;
import org.apache.flink.table.connector.format.EncodingFormat;
import org.apache.flink.table.connector.sink.DynamicTableSink;
import org.apache.flink.table.connector.sink.SinkFunctionProvider;
import org.apache.flink.table.data.RowData;
import org.apache.flink.table.types.DataType;

/**
 * @author lijiaqi
 */
public class OpenGaussDynamicTableSink implements DynamicTableSink {

    private final JdbcOptions jdbcOptions;
    private final EncodingFormat<SerializationSchema<RowData>> encodingFormat;
    private final DataType dataType;

    public OpenGaussDynamicTableSink(JdbcOptions jdbcOptions, EncodingFormat<SerializationSchema<RowData>> encodingFormat, DataType dataType) {
        this.jdbcOptions = jdbcOptions;
        this.encodingFormat = encodingFormat;
        this.dataType = dataType;
    }

    @Override
    public ChangelogMode getChangelogMode(ChangelogMode requestedMode) {
        return requestedMode;
    }

    @Override
    public SinkRuntimeProvider getSinkRuntimeProvider(Context context) {
        System.out.println("SinkRuntimeProvider");
        System.out.println(dataType);

//        SerializationSchema<RowData> serializationSchema = encodingFormat.createRuntimeEncoder(context, dataType);
        OpenGaussSinkFunction gbasedbtSinkFunction = new OpenGaussSinkFunction(jdbcOptions,dataType);
        return SinkFunctionProvider.of(gbasedbtSinkFunction);
    }

    @Override
    public DynamicTableSink copy() {
        return new OpenGaussDynamicTableSink(jdbcOptions, encodingFormat, dataType);
    }

    @Override
    public String asSummaryString() {
        return "OpenGauss Table Sink";
    }

}

构建OpenGaussDynamicTableFactory

package name.lijiaqi.table;


import name.lijiaqi.dialect.OpenGaussDialect;
import org.apache.flink.configuration.ConfigOption;
import org.apache.flink.configuration.ConfigOptions;
import org.apache.flink.configuration.ReadableConfig;
import org.apache.flink.connector.jdbc.internal.options.JdbcOptions;
import org.apache.flink.table.api.TableSchema;
import org.apache.flink.table.connector.sink.DynamicTableSink;
import org.apache.flink.table.connector.source.DynamicTableSource;
import org.apache.flink.table.factories.DynamicTableSinkFactory;
import org.apache.flink.table.factories.DynamicTableSourceFactory;
import org.apache.flink.table.factories.FactoryUtil;
import org.apache.flink.table.types.DataType;
import org.apache.flink.table.utils.TableSchemaUtils;

import java.util.HashSet;
import java.util.Set;

/**
 * @author lijiaqi
 */
public class OpenGaussDynamicTableFactory implements DynamicTableSourceFactory, DynamicTableSinkFactory {

    public static final String IDENTIFIER = "opengauss";

    private static final String DRIVER_NAME = "org.opengauss.Driver";

    public static final ConfigOption<String> URL = ConfigOptions
            .key("url")
            .stringType()
            .noDefaultValue()
            .withDescription("the jdbc database url.");

    public static final ConfigOption<String> DRIVER = ConfigOptions
            .key("driver")
            .stringType()
            .defaultValue(DRIVER_NAME)
            .withDescription("the jdbc driver.");



    public static final ConfigOption<String> TABLE_NAME = ConfigOptions
            .key("table-name")
            .stringType()
            .noDefaultValue()
            .withDescription("the jdbc table name.");

    public static final ConfigOption<String> USERNAME = ConfigOptions
            .key("username")
            .stringType()
            .noDefaultValue()
            .withDescription("the jdbc user name.");

    public static final ConfigOption<String> PASSWORD = ConfigOptions
            .key("password")
            .stringType()
            .noDefaultValue()
            .withDescription("the jdbc password.");

//    public static final ConfigOption<String> FORMAT = ConfigOptions
//            .key("format")
//            .stringType()
//            .noDefaultValue()
//            .withDescription("the format.");

    @Override
    public String factoryIdentifier() {
        return IDENTIFIER;
    }

    @Override
    public Set<ConfigOption<?>> requiredOptions() {
        Set<ConfigOption<?>> requiredOptions = new HashSet<>();
        requiredOptions.add(URL);
        requiredOptions.add(TABLE_NAME);
        requiredOptions.add(USERNAME);
        requiredOptions.add(PASSWORD);
//        requiredOptions.add(FORMAT);
        return requiredOptions;
    }

    @Override
    public Set<ConfigOption<?>> optionalOptions() {
        return new HashSet<>();
    }

    @Override
    public DynamicTableSource createDynamicTableSource(Context context) {

        final FactoryUtil.TableFactoryHelper helper = FactoryUtil.createTableFactoryHelper(this, context);

        final ReadableConfig config = helper.getOptions();

        helper.validate();

        JdbcOptions jdbcOptions = getJdbcOptions(config);

        TableSchema physicalSchema = TableSchemaUtils.getPhysicalSchema(context.getCatalogTable().getSchema());

        return new OpenGaussDynamicTableSource(jdbcOptions, physicalSchema);

    }

    @Override
    public DynamicTableSink createDynamicTableSink(Context context) {

        final FactoryUtil.TableFactoryHelper helper = FactoryUtil.createTableFactoryHelper(this, context);

//        final EncodingFormat<SerializationSchema<RowData>> encodingFormat = helper.discoverEncodingFormat(
//                SerializationFormatFactory.class,
//                FactoryUtil.FORMAT);

        final ReadableConfig config = helper.getOptions();

        // validate all options
        helper.validate();

        // get the validated options
        JdbcOptions jdbcOptions = getJdbcOptions(config);

        // derive the produced data type (excluding computed columns) from the catalog table
        final DataType dataType = context.getCatalogTable().getSchema().toPhysicalRowDataType();

        // table sink
        return new OpenGaussDynamicTableSink(jdbcOptions, null, dataType);
    }

    private JdbcOptions getJdbcOptions(ReadableConfig readableConfig) {
        final String url = readableConfig.get(URL);
        final JdbcOptions.Builder builder = JdbcOptions.builder()
                .setDriverName(DRIVER_NAME)
                .setDBUrl(url)
                .setTableName(readableConfig.get(TABLE_NAME))
                .setDialect(new OpenGaussDialect());

        readableConfig.getOptional(USERNAME).ifPresent(builder::setUsername);
        readableConfig.getOptional(PASSWORD).ifPresent(builder::setPassword);
        return builder.build();
    }

}

接下来通过SPI注册动态表：创建文件resources\\META-INF\\services\\org.apache.flink.table.factories.Factory内容注册为name.lijiaqi.table.OpenGaussDynamicTableFactory

至此，我们的Flink connector 就构建完成，接下来，我们要使用其，来完成一个真正的项目。

CDC实战

下面是项目的整体架构图，我们通过flink cdc 从mysql获取变更数据，然后通过 flink sql 将数据 sink 到 opengauss里

接下来，我们看一下如何通过Flink SQL实现CDC ，只需3条SQL语句即可。

创建数据源表

       // 数据源表
       String sourceDDL =
               "CREATE TABLE mysql_binlog (\\n" +
                       " id INT NOT NULL,\\n" +
                       " name STRING,\\n" +
                       " description STRING\\n" +
                       ") WITH (\\n" +
                       " 'connector' = 'mysql-cdc',\\n" +
                       " 'hostname' = 'localhost',\\n" +
                       " 'port' = '3306',\\n" +
                       " 'username' = 'flinkcdc',\\n" +
                       " 'password' = '123456',\\n" +
                       " 'database-name' = 'test',\\n" +
                       " 'table-name' = 'test_cdc'\\n" +
                       ")";

创建输出表，输出到opengauss ，这里 connector设置成opengauss

        String url = "jdbc:opengauss://172.27.71.161:26000/postgres";
        String userName = "jacky";
        String password = "123456";
        String gbasedbtSinkTable = "t1";
        // 输出目标表
        String sinkDDL =
                "CREATE TABLE test_cdc_sink (\\n" +
                        " id INT NOT NULL,\\n" +
                        " name STRING,\\n" +
                        " description STRING,\\n" +
                        " PRIMARY KEY (id) NOT ENFORCED \\n " +
                        ") WITH (\\n" +
                        " 'connector' = 'opengauss',\\n" +
//                       " 'driver' = 'com.gbasedbt.jdbc.Driver',\\n" +
                        " 'url' = '" + url + "',\\n" +
                        " 'username' = '" + userName + "',\\n" +
                        " 'password' = '" + password + "',\\n" +
                        " 'table-name' = '" + gbasedbtSinkTable + "' \\n" +
                        ")";

这里我们直接将数据汇入

       String transformSQL =
               "insert into test_cdc_sink select * from mysql_binlog";

完整参考代码

package name.lijiaqi;

import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.table.api.EnvironmentSettings;
import org.apache.flink.table.api.SqlDialect;
import org.apache.flink.table.api.TableResult;
import org.apache.flink.table.api.bridge.java.StreamTableEnvironment;

public class MysqlToOpenGaussMain {
    public static void main(String[] args) throws Exception {
        EnvironmentSettings fsSettings = EnvironmentSettings.newInstance()
                .useBlinkPlanner()
                .inStreamingMode()
                .build();
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setParallelism(1);
        StreamTableEnvironment tableEnv = StreamTableEnvironment.create(env, fsSettings);



        tableEnv.getConfig().setSqlDialect(SqlDialect.DEFAULT);


        // 数据源表
        String sourceDDL =
                "CREATE TABLE mysql_binlog (\\n" +
                        " id INT NOT NULL,\\n" +
                        " name STRING,\\n" +
                        " description STRING,\\n" +
                        " PRIMARY KEY (id) NOT ENFORCED" +
                        ") WITH (\\n" +
                        " 'connector' = 'mysql-cdc',\\n" +
                        " 'hostname' = 'localhost',\\n" +
                        " 'port' = '3306',\\n" +
                        " 'username' = 'flinkcdc',\\n" +
                        " 'password' = '123456',\\n" +
                        " 'database-name' = 'test',\\n" +
                        " 'table-name' = 'test_cdc'\\n" +
                        ")";


        String url = "jdbc:opengauss://172.27.71.161:26000/postgres";
        String userName = "jacky";
        String password = "123456";
        String gbasedbtSinkTable = "t1";
        // 输出目标表
        String sinkDDL =
                "CREATE TABLE test_cdc_sink (\\n" +
                        " id INT NOT NULL,\\n" +
                        " name STRING,\\n" +
                        " description STRING,\\n" +
                        " PRIMARY KEY (id) NOT ENFORCED \\n " +
                        ") WITH (\\n" +
                        " 'connector' = 'opengauss',\\n" +
//                       " 'driver' = 'com.gbasedbt.jdbc.Driver',\\n" +
                        " 'url' = '" + url + "',\\n" +
                        " 'username' = '" + userName + "',\\n" +
                        " 'password' = '" + password + "',\\n" +
                        " 'table-name' = '" + gbasedbtSinkTable + "' \\n" +
                        ")";

        String transformSQL =
                "insert into test_cdc_sink select * from mysql_binlog";

        tableEnv.executeSql(sourceDDL);
        tableEnv.executeSql(sinkDDL);
        TableResult result = tableEnv.executeSql(transformSQL);

        result.print();
        env.execute("sync-flink-cdc");
    }

}

运行结果