第1章 debezium概述

1.1 debezium是什么

Debezium 是一组分布式服务,用于捕获数据库中的更改,以便您的应用程序可以查看这些更改并对其做出响应。 Debezium 在更改事件流中记录每个数据库表中的所有行级更改,应用程序只需读取这些流以查看更改事件发生的相同顺序。

1.2 debezium的依赖

Debezium 构建在 Apache Kafka 之上,并提供与 Kafka Connect 兼容的连接器,用于监控特定的数据库管理系统。 Debezium 在 Kafka 日志中记录数据更改的历史记录,从您的应用程序使用它们的位置。 这使您的应用程序可以轻松正确且完整地使用所有事件。 即使您的应用程序意外停止,它也不会错过任何内容:当应用程序重新启动时,它将继续使用它停止的事件。

第2章 环境搭建

2.1 所需软件及测试版本

jdk-8u211-linux-x64.tar.gz

apache-zookeeper-3.5.7-bin.tar.gz

kafka_2.11-2.4.0.tgz

debezium-connector-oracle-1.7.0.Final-plugin.tar.gz

instantclient-basic-linux.x64-21.3.0.0.0.zip (oracle客户端)

oracle 11g R2

2.2 数据库准备

(1)设置Oracle LogMiner所需的配置

[root@tDataAnalysis01 config]# su - oracle
[oracle@tDataAnalysis01 config]$ sqlplus / as sysdba
SQL> alter system set db_recovery_file_dest_size = 10G;
SQL> shutdown immediate;
SQL> startup mount;
SQL> alter database archivelog;
SQL> alter database open;

(2)保证Database log mode显示为Archive Mode

SQL> archive log list

(3)启用补充日志记录,(所有表都得执行,表里须至少一条数据)

SQL> ALTER DATABASE add SUPPLEMENTAL LOG DATA ;
SQL> ALTER TABLE 库.表 ADD SUPPLEMENTAL LOG DATA (ALL) COLUMNS;

Debezium Oracle连接器要求使用特定权限设置用户账号,以便连接器捕获更多事件。

CREATE TABLESPACE logminer_tbs DATAFILE '/opt/oracle/oradata/ORCLCDB/logminer_tbs.dbf'
SIZE 25M REUSE AUTOEXTEND ON MAXSIZE UNLIMITED;

CREATE USER debezium IDENTIFIED BY 123456
    DEFAULT TABLESPACE TEST0
    QUOTA UNLIMITED ON TEST0;
 
  GRANT CREATE SESSION TO debezium ;
  GRANT SELECT ON V_$DATABASE to debezium ;
  GRANT FLASHBACK ANY TABLE TO debezium ;
  GRANT SELECT ANY TABLE TO debezium ;
  GRANT SELECT_CATALOG_ROLE TO debezium ;
  GRANT EXECUTE_CATALOG_ROLE TO debezium ;
  GRANT SELECT ANY TRANSACTION TO debezium ;

  GRANT CREATE TABLE TO debezium ;
  GRANT LOCK ANY TABLE TO debezium ;
  GRANT ALTER ANY TABLE TO debezium ;
  GRANT CREATE SEQUENCE TO debezium ;
 
  GRANT EXECUTE ON DBMS_LOGMNR TO debezium ;
  GRANT EXECUTE ON DBMS_LOGMNR_D TO debezium ;
 
  GRANT SELECT ON V_$LOG TO debezium ;
  GRANT SELECT ON V_$LOG_HISTORY TO debezium ;
  GRANT SELECT ON V_$LOGMNR_LOGS TO debezium ;
  GRANT SELECT ON V_$LOGMNR_CONTENTS TO debezium ;
  GRANT SELECT ON V_$LOGMNR_PARAMETERS TO debezium ;
  GRANT SELECT ON V_$LOGFILE TO debezium ;
  GRANT SELECT ON V_$ARCHIVED_LOG TO debezium ;
  GRANT SELECT ON V_$ARCHIVE_DEST_STATUS TO debezium ;
 
  exit;

2.3部署连接器

(1)进入到oracle客户端目录中,将xstreams.jar和ojbc8.jar复制到kafka lib下,并分发给02,03两台机器。

[root@tDataAnalysis01 ~]# cd /opt/oracle/ instantclient_21_3/
[root@tDataAnalysis01 instantclient_21_3]# cp xstreams.jar /opt/module/kafka_2.11-2.4.0/libs/
[root@tDataAnalysis01 instantclient_21_3]# cp ojdbc8.jar /opt/module/kafka_2.11-2.4.0/libs/
[root@tDataAnalysis01 instantclient_21_3]# scp xstreams.jar tDataAnalysis02:/opt/module/kafka_2.11-2.4.0/libs/
[root@tDataAnalysis01 instantclient_21_3]# scp ojdbc8.jar tDataAnalysis02:/opt/module/kafka_2.11-2.4.0/libs/
[root@tDataAnalysis01 instantclient_21_3]# scp xstreams.jar tDataAnalysis03:/opt/module/kafka_2.11-2.4.0/libs/
[root@tDataAnalysis01 instantclient_21_3]# scp ojdbc8.jar tDataAnalysis03:/opt/module/kafka_2.11-2.4.0/libs/

(2)增加LD_LIBRARY_PATH环境变量,指向oracle client路径(每个节点都配置)

[root@tDataAnalysis01 instantclient_21_3]# vim /etc/profile
export LD_LIBRARY_PATH=/opt/oracle/instantclient_21_3
[root@tDataAnalysis01 instantclient_21_3]# source /etc/profile

(3)创建kafka plugins文件夹

[root@tDataAnalysis01 software]# mkdir -p /usr/local/share/kafka/plugins/

(4)上传debezium-connector-oracle-1.7.0.Final-plugin.tar.gz
,解压到kafka plugins目录

[root@tDataAnalysis01 software]# tar -zxvf debezium-connector-oracle-1.7.0.Final-plugin.tar.gz -C /usr/local/share/kafka/plugins/

(5)再次修改kafka下的connect-distributed.properties文件,添加plugins路径

[root@tDataAnalysis01 software]# cd /opt/module/kafka_2.11-2.4.0/config/
[root@tDataAnalysis01 config]# vim connect-distributed.properties 
plugin.path=/usr/local/share/kafka/plugins/

(6)分发plugins文件夹,分发connect-distributed.properties

[root@tDataAnalysis01 config]# scp -r /usr/local/share/kafka/  tDataAnalysis02:/usr/local/share/
[root@tDataAnalysis01 config]# scp -r /usr/local/share/kafka/  tDataAnalysis03:/usr/local/share/
[root@tDataAnalysis01 config]# scp connect-distributed.properties  tDataAnalysis02:/opt/module/kafka_2.11-2.4.0/config/
[root@tDataAnalysis01 config]# scp connect-distributed.properties  tDataAnalysis03:/opt/module/kafka_2.11-2.4.0/config/

(7)启动kafka connector

[root@tDataAnalysis01 kafka_2.11-2.4.0]# bin/connect-distributed.sh -daemon config/connect-distributed.properties
[root@tDataAnalysis02 kafka_2.11-2.4.0]# bin/connect-distributed.sh -daemon config/connect-distributed.properties
[root@tDataAnalysis03 kafka_2.11-2.4.0]# bin/connect-distributed.sh -daemon config/connect-distributed.properties

(8)查看进程

25010 Kafka
25555 ConnectDistributed

第3章启动Debezium

3.1 测试任务启动

curl -H "Content-Type: application/json" -X POST -d  '{
       "name" : "oracle-connector",
       "config" : {
           "connector.class" : "io.debezium.connector.oracle.OracleConnector",
           "database.hostname" : "xxx.xxx.16.53",
           "database.port" : "1522",
           "database.user" : "debezium",
           "database.password" : "123456",
           "database.dbname": "ORCL",
           "database.server.name" : "orcl6",
           "database.history.kafka.topic" : "schema-changes.inventory5",
           "database.history.kafka.bootstrap.servers" : "tDataAnalysis01:9092,tDataAnalysis02:9092,tDataAnalysis03:9092",
           "snapshot.mode" : "initial",
           "schema.include.list" : "c##HAHA123",
           "table.include.list" : "c##HAHA123.CIRCLE",
           "database.oracle.version": "11", 
           "database.tablename.case.insensitive": "false",
           "decimal.handling.mode" : "string"
 }
}' http://tDataAnalysis01:8083/connectors

测试任务这里我使用了初始化快照

查看kafka topic

bin/kafka-topics.sh --bootstrap-server tDataAnalysis01:9092 --list
FlinkError
__consumer_offsets
connect-oracle-configs
connect-oracle-offsets
connect-oracle-status
flumeError
myTest
orcl6.C__HAHA123.CIRCLE
schema-changes.inventory5

查看connect状态

curl -H "Accept:application/json" tDataAnalysis01:8083/connectors/
["oracle-connector"][root@tDataAnalysis02 kafka]#

插入数据后

消费topic

bin/kafka-console-consumer.sh --bootstrap-server tDataAnalysis01:9092 --topic orcl6.C__HAHA123.CIRCLE --from-beginning

数据格式

{
    "schema": {
        "type": "struct", 
        "fields": [
            {
                "type": "struct", 
                "fields": [
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "SERIALNO"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "FILEPATH"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "PARTID"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "STAFFNO"
                    }, 
                    {
                        "type": "int64", 
                        "optional": false, 
                        "name": "io.debezium.time.Timestamp", 
                        "version": 1, 
                        "field": "RECORDTIME"
                    }
                ], 
                "optional": true, 
                "name": "orcl6.C__HAHA123.CIRCLE.Value", 
                "field": "before"
            }, 
            {
                "type": "struct", 
                "fields": [
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "SERIALNO"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "FILEPATH"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "PARTID"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "STAFFNO"
                    }, 
                    {
                        "type": "int64", 
                        "optional": false, 
                        "name": "io.debezium.time.Timestamp", 
                        "version": 1, 
                        "field": "RECORDTIME"
                    }
                ], 
                "optional": true, 
                "name": "orcl6.C__HAHA123.CIRCLE.Value", 
                "field": "after"
            }, 
            {
                "type": "struct", 
                "fields": [
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "version"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "connector"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "name"
                    }, 
                    {
                        "type": "int64", 
                        "optional": false, 
                        "field": "ts_ms"
                    }, 
                    {
                        "type": "string", 
                        "optional": true, 
                        "name": "io.debezium.data.Enum", 
                        "version": 1, 
                        "parameters": {
                            "allowed": "true,last,false"
                        }, 
                        "default": "false", 
                        "field": "snapshot"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "db"
                    }, 
                    {
                        "type": "string", 
                        "optional": true, 
                        "field": "sequence"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "schema"
                    }, 
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "table"
                    }, 
                    {
                        "type": "string", 
                        "optional": true, 
                        "field": "txId"
                    }, 
                    {
                        "type": "string", 
                        "optional": true, 
                        "field": "scn"
                    }, 
                    {
                        "type": "string", 
                        "optional": true, 
                        "field": "commit_scn"
                    }, 
                    {
                        "type": "string", 
                        "optional": true, 
                        "field": "lcr_position"
                    }
                ], 
                "optional": false, 
                "name": "io.debezium.connector.oracle.Source", 
                "field": "source"
            }, 
            {
                "type": "string", 
                "optional": false, 
                "field": "op"
            }, 
            {
                "type": "int64", 
                "optional": true, 
                "field": "ts_ms"
            }, 
            {
                "type": "struct", 
                "fields": [
                    {
                        "type": "string", 
                        "optional": false, 
                        "field": "id"
                    }, 
                    {
                        "type": "int64", 
                        "optional": false, 
                        "field": "total_order"
                    }, 
                    {
                        "type": "int64", 
                        "optional": false, 
                        "field": "data_collection_order"
                    }
                ], 
                "optional": true, 
                "field": "transaction"
            }
        ], 
        "optional": false, 
        "name": "orcl6.C__HAHA123.CIRCLE.Envelope"
    }, 
    "payload": {
        "before": null, 
        "after": {
            "SERIALNO": "1209", 
            "FILEPATH": "vqrouj", 
            "PARTID": "5606", 
            "STAFFNO": "YTCZ060002", 
            "RECORDTIME": 1329061915000
        }, 
        "source": {
            "version": "1.7.0.Final", 
            "connector": "oracle", 
            "name": "orcl6", 
            "ts_ms": 1634660571000, 
            "snapshot": "false", 
            "db": "ORCL", 
            "sequence": null, 
            "schema": "C##HAHA123", 
            "table": "CIRCLE", 
            "txId": "05000300282e0000", 
            "scn": "19202446", 
            "commit_scn": "19202478", 
            "lcr_position": null
        }, 
        "op": "c", 
        "ts_ms": 1634631840937, 
        "transaction": null
    }
}

删除connect

curl -v -X DELETE http://tDataAnalysis01:8083/connectors/oracle-connector

访问 8083端口可查看 connect状态

tDataAnalysis03:8083/connectors/oracle-connector/status

3.2 Oracle归档日志过多解决(磁盘很大可以忽略)

测试中因为日志写入过快问题导致磁盘被写满,使得Oracle数据库极不稳定

日志不允许物理删除,应通过rman target /

执行

delete noprompt archivelog until time 'sysdate-1/24';

脚本处理

# 由于连接器是依赖于Oracle的归档日志做的实时同步
# 存在归档日志撑爆磁盘的问题
#! /bin/bash
exec >> /xxx/oracle/oracle_log/del_arch`date +%F-%H`.log #记录脚本日志
$ORACLE_HOME/bin/rman target / <<EOF
#检查归档日志
crosscheck archivelog all;
#删除所有过期日志
delete noprompt expired archivelog all;
#删除一个小时前的归档日志
delete noprompt archivelog until time 'sysdate-1/24';
exit;
EOF

3.3 新增表问题

3.3.1 第一种解决方案

第一次安装Debezium,可以监控到数据库中表的初始表,但是后续再建表并且增加SUPPLEMENTAL LOG DATA,Connector是无法捕获到的。
解决方案:
再提交一个新的connecotr,用来监控后续创建的表

curl -H "Content-Type: application/json" -X POST -d  '{
       "name" : "oracle-connector2",
       "config" : {
           "connector.class" : "io.debezium.connector.oracle.OracleConnector",
           "database.hostname" : "xxx.xxx.16.53",
           "database.port" : "1522",
           "database.user" : "debezium",
           "database.password" : "123456",
           "database.dbname": "ORCL",
           "database.server.name" : "orcl7",
           "database.history.kafka.topic" : "schema-changes.inventory6",
           "database.history.kafka.bootstrap.servers" : "tDataAnalysis01:9092,tDataAnalysis02:9092,tDataAnalysis03:9092",
           "snapshot.mode" : "initial",
           "schema.include.list" : "c##HAHA123",
           "table.include.list" : "c##HAHA123.CIRCLE2",
           "database.oracle.version": "11", 
           "database.tablename.case.insensitive": "false",
           "decimal.handling.mode" : "string"
 }
}' http://tDataAnalysis01:8083/connectors

那么可以使用参数table.include.list指定所需的监控的表,或者使用table.exclude.list指定不需要监控的表,二者只能选一个使用(这个方案实测通过,可以正常收数据)

查看连接器状态:

curl -H "Accept:application/json" tDataAnalysis01:8083/connectors/
["oracle-connector2","oracle-connector"][root@tDataAnalysis02 kafka]#
3.3.2第二种解决方案

*最新测试 根据社区回复

在这里插入图片描述
新表可通过停止原有任务,修改配置参数,重新启动任务,旧表数据不受影响,底层存有偏移量,数据同步可以追溯 快照方式为 schema_only

建议: 我们的oracle数据库本身数据量存储量应该是不小的,我在测试中为方便快照为初始化,生产中建议schema_only 为增量读取以避免任务长时间运行造成失败。

3.4 注意事项

3.4.1配置 Debezium 主题

Debezium 使用(通过 Kafka Connect 或直接)多个主题来存储数据。 这些主题必须由管理员或 Kafka 本身通过启用主题的自动创建来创建。
有一些适用于主题的限制和建议:
数据库历史主题(适用于 MySQL 和 SQL Server 的 Debezium 连接器)
无限(或非常长)保留(无压缩!)
生产的复制因子至少为 3
单分区

3.4.2其他主题

可选地,启用日志压缩(如果您希望只保留给定记录的最后一个更改事件); 在这种情况下,应配置 Apache Kafka 中的 min.compaction.lag.ms 和 delete.retention.ms 主题级别设置,以便消费者有足够的时间接收所有事件并删除标记; 具体来说,这些值应大于您预期的接收器连接器的最大停机时间,例如 更新它们时
在生产中复制

3.4.3单分区

您可以放宽单分区规则,但您的应用程序必须处理数据库中不同行的乱序事件(单行的事件仍然是完全有序的)。 如果使用多个分区,Kafka默认会通过对key进行hash来确定分区。 其他分区策略需要使用 SMT 为每个记录设置分区号。
有关可自定义的主题自动创建(自 Kafka Connect 2.6.0 起可用),请参阅自定义主题自动创建

官网对于数据库历史主题虽只提到MySQL,SQL server,但是实际测试中如果schema-changes.inventory5 主题被误删,connect将报错,只有将ddl语句重新插入进去,才可以恢复正常,且oracle 连接器快照目前只支持schema_only,initial 。临时快照和增量快照尚在孵化中

3.4.4关于对oracle 数据库压力的问题

社区大佬的回复

3.5 错误处理

如出现不能识别oracle版本或其他错误,请前往xx01(post请求后缀为01)节点所在Kafka logs目录
执行

tail -200 connectDistributed.out

查看具体日志

3.5.1不能识别jdbc驱动?

先去看oracle配置,确保host,port,sid等无误,另外判断ojdbc8.jar是否被识别到,可以尝试重启Kafka 以及Kafka connect 或直接新装一套Kafka集群进行验证。

3.5.2 Kafka以及Kafka connect重启问题

Kafka(暴力杀死,重启会有冲突) 不允许kill -9 杀掉进程 应使用命令停止,Kafka通常关闭较慢,确认集群Kafka关闭后再重启 各个节点执行停止或重启,cdh版在控制台启停即可。

/opt/module/kafka/bin/kafka-server-stop.sh -daemon /opt/module/kafka/config/server.properties

如需重启 Kafka connect服务
Kafka connect kill -9 杀掉所有节点即可
启动命令

/opt/module/kafka/bin/connect-distributed.sh -daemon /opt/module/kafka/config/connect-distributed.properties 

参考

https://debezium.io/documentation/reference/1.7/connectors/oracle.html#oracle-overview

尚硅谷大数据技术之Oracle实时同步方案.doc (debezium to oracle 19 c)

https://blog.51cto.com/tiany/1552783

https://www.codeleading.com/article/64945986138/

https://my.oschina.net/dacoolbaby/blog/4320036

https://my.oschina.net/dacoolbaby/blog/3191882

https://www.cnblogs.com/yaowentao/p/14944739.html

Logo

华为开发者空间,是为全球开发者打造的专属开发空间,汇聚了华为优质开发资源及工具,致力于让每一位开发者拥有一台云主机,基于华为根生态开发、创新。

更多推荐