客服微信
TDSQL 备份故障处理案例
TDSQL 备份意义：
有人说分布数据库都是多副本的，没有必要进行备份。我觉得这话只对了一半，首先分布式数据库是多副本的没错。但还是有必要进行备份的。 例如数据误删，drop table ,drop database 操作;或者需要进行历史数据的分析;又或者多重灾难，导致主备数据全部损毁。可能这些机率很小，得我们的数据经不起这样的损毁。所以不管是传统集中式数据库，还是现在比较流行的分布式数据库。为了数据安全，理应进行备份。
TDSQL备份介绍：
TDSQL支持物理备份和逻辑备份，它会自动从一个副本进行异步实时备份，同时能守binlog进行增量备份。利用最近一天的镜像，结合期间的binlog即完成指定时间点的恢复。 物理备份是其于xtrabackup在底层拷贝一致性数据文件+BINLOG位置。逻辑备份基于mydumper select获取一致性数据＋BINLOG位置。其核心原理就是通过一致性全量数据+BINLOG位置。
TDSQL 备份失败案例：
通过TDSQL的物理备份对配置库 4001进行备份的时候。
提示: Error occurred,see mysqlagent log for detail.错误如下：
查看备份节点的日志：
nohup]# pwd/data/tdsql_run/4001/mysqlagent/log/nohupnohup]# lltotal 481 tdsql users 3 Dec 12 15:53 coldbackupbinlog_result_40011 tdsql users 34077 Dec 12 15:49 coldbackupimage_4001_2020-12-121 tdsql users 3 Dec 12 15:27 coldbackup_result_40011 tdsql users 3 Dec 12 15:49 manualBackupResult_4001_0000000000.logtail -100f coldbackupimage_4001_2020-12-12(errno: 2, cannot open shared object file: No such file or directory)innobackupex: Error writing file 'UNOPENED' (Errcode: 32 - Broken pipe)failed.xtrabackup: Error: write to logfile failedinnobackupex: Error writing file 'UNOPENED' (Errcode: 32 - Broken pipe)xtrabackup: Error: xtrabackup_copy_logfile() failed.real 0m0.459suser 0m0.420ssys 0m0.063spipe:1 141 0 1,result:-1-1 2020-12-12 15:49:23ERROR: JAVA_HOME is not set and could not be found.
#处理：
[][]java version "1.8.0_152"Java(TM) SE Runtime Environment (build 1.8.0_152-b16)Java HotSpot(TM) 64-Bit Server VM (build 25.152-b16, mixed mode)
#以tdsql 用户重启验证
[root@huyidb03 nohup]# su - tdsql[tdsql@huyidb03 nohup]# cd /data/tdsql_run/4001/mysqlagent/bin[tdsql@huyidb03 bin]# ./restartreport_cgroup.sh ../conf/mysqlagent_4001.xmlstop mysqlreport successstart mysqlreport success
#而此时日志继续报错，无法连接hadoop
details see: http://wiki.apache.org/hadoop/ConnectionRefusedls: Call From huyidb03/10.85.10.53 to huyidb01:9002 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefusedput: Call From huyidb03/10.85.10.53 to huyidb01:9002 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefusedappendToFile: Call From huyidb03/10.85.10.53 to huyidb01:9002 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefusedput: Call From huyidb03/10.85.10.53 to huyidb01:9002 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused
#hadoop 节点去检查
[tdsql@huyidb01 ~]$ hadoop fs -ls /ls: Call From huyidb01/10.85.10.51 to huyidb01:9002 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused
#再次检查hadoop：
[tdsql@huyidb01 ~]$ hadoop fs -ls /Found 1 itemsdrwxr-xr-x - tdsql supergroup 0 2020-12-12 12:37 /tdsqlbackup
#备份节点检查hadoop挂载情况：
[]Found 1 itemsdrwxr-xr-x - tdsql supergroup 0 2020-12-12 12:37 /tdsqlbackup[]
#再次在备份节点上发起备份OK
原因分析：
初步分析是因为我们先安装的4001的实例和agent，然后才安装的hadoop，而hadoop安装好后没有重启4001的agent 让其配置生效。才导至在备份4001实例时失败，而其它实例正常。
