其他分享
首页 > 其他分享> > 11g 逻辑备库简单故障处理

11g 逻辑备库简单故障处理

作者:互联网

模拟同步失败  

备库创建唯一性索引

SQL> create unique index idx_scott_t_1 on scott.t(user_id);  

Index created.  


主库插入重复数据  

SQL> insert into scott.t select * from scott.t;


36 rows created.  

SQL> commit;  

Commit complete.  



备库自动停止同步  

Mon Sep 29 17:22:32 2014  
LOGSTDBY: SQL Apply about to stop with ORA-1  
LOGSTDBY: SQL Apply finished logging error information  
LOGSTDBY Apply process AS02 server id=2 pid=41 OS id=2535 stopped  
Errors in file    /u01/app/oracle/diag/rdbms/orcl2/orcl/trace/orcl_as02_2535.trc:  
ORA-00001: unique constraint (SYS.IDX_SCOTT_T_1) violated  
Mon Sep 29 17:22:36 2014  
Errors in file /u01/app/oracle/diag/rdbms/orcl2/orcl/trace/orcl_lsp0_2507.trc:  
ORA-26808: Apply process AS02 died unexpectedly.  
ORA-00001: unique constraint (SYS.IDX_SCOTT_T_1) violated  
Mon Sep 29 17:22:36 2014  
LOGSTDBY Analyzer process AS00 server id=0 pid=39 OS id=2531 stopped  
Mon Sep 29 17:22:36 2014  
LOGSTDBY Apply process AS01 server id=1 pid=40 OS id=2533 stopped  
Mon Sep 29 17:22:36 2014  
LOGSTDBY Apply process AS03 server id=3 pid=42 OS id=2537 stopped  
Mon Sep 29 17:22:36 2014  
LOGSTDBY Apply process AS04 server id=4 pid=43 OS id=2539 stopped  
Mon Sep 29 17:22:36 2014  
LOGSTDBY Apply process AS05 server id=5 pid=44 OS id=2541 stopped  
Mon Sep 29 17:22:40 2014  
LOGMINER: session#=1 (Logical_Standby$), preparer MS02 pid=36 OS id=2518 sid=46 stopped  
Mon Sep 29 17:22:40 2014  
LOGMINER: session#=1 (Logical_Standby$), reader MS00 pid=34 OS id=2514 sid=44 stopped  
Mon Sep 29 17:22:40 2014  
LOGMINER: session#=1 (Logical_Standby$), builder MS01 pid=35 OS id=2516 sid=34 stopped  
LOGSTDBY status: ORA-16222: automatic Logical Standby retry of last action  
LOGMINER: Parameters summary for session# = 1  
LOGMINER: Number of processes = 3, Transaction Chunk Size = 201  
LOGMINER: Memory Size = 30M, Checkpoint interval = 150M  
LOGMINER: SpillScn 1114702, ResetLogScn 995548  
LOGMINER: summary for session# = 1  
LOGMINER: StartScn: 0 (0x0000.00000000)  
LOGMINER: EndScn: 0 (0x0000.00000000)  
LOGMINER: HighConsumedScn: 1114646 (0x0000.00110216)  
LOGMINER: session_flag: 0x1  
LOGMINER: Read buffers: 16  
Mon Sep 29 17:22:43 2014  
LOGMINER: session#=1 (Logical_Standby$), reader MS00 pid=34 OS id=2554 sid=34 started  
Mon Sep 29 17:22:43 2014  
LOGMINER: session#=1 (Logical_Standby$), builder MS01 pid=35 OS id=2556 sid=44 started  
Mon Sep 29 17:22:43 2014  
LOGMINER: session#=1 (Logical_Standby$), preparer MS02 pid=36 OS id=2558 sid=45 started  
LOGMINER: Turning ON Log Auto Delete  
Mon Sep 29 17:22:45 2014  
LOGSTDBY Analyzer process AS00 started with server id=0 pid=40 OS id=2560  
Mon Sep 29 17:22:45 2014  
LOGSTDBY Apply process AS01 started with server id=1 pid=41 OS id=2562  
Mon Sep 29 17:22:45 2014  
LOGSTDBY Apply process AS04 started with server id=4 pid=44 OS id=2568  
Mon Sep 29 17:22:45 2014  
Mon Sep 29 17:22:45 2014  
LOGSTDBY Apply process AS05 started with server id=5 pid=45 OS id=2570  
Mon Sep 29 17:22:45 2014  
LOGSTDBY Apply process AS03 started with server id=3 pid=43 OS id=2566  
LOGSTDBY Apply process AS02 started with server id=2 pid=42 OS id=2564  
LOGMINER: Begin mining logfile for session 1 thread 1 sequence 42, /u01/app/oracle/fast_recovery_area/ORCL2/onlinelog/o1_mf_4_b2l8vov0_.log  
LOGSTDBY: SQL Apply about to stop with ORA-1  
LOGSTDBY: SQL Apply finished logging error information  
LOGSTDBY Apply process AS01 server id=1 pid=41 OS id=2562 stopped  
Errors in file /u01/app/oracle/diag/rdbms/orcl2/orcl/trace/orcl_lsp0_2507.trc:  
ORA-26808: Apply process AS01 died unexpectedly.  
ORA-00001: unique constraint (SYS.IDX_SCOTT_T_1) violated  
LOGSTDBY Analyzer process AS00 server id=0 pid=40 OS id=2560 stopped  
LOGSTDBY Apply process AS02 server id=2 pid=42 OS id=2564 stopped  
LOGSTDBY Apply process AS03 server id=3 pid=43 OS id=2566 stopped  
LOGSTDBY Apply process AS04 server id=4 pid=44 OS id=2568 stopped  
LOGSTDBY Apply process AS05 server id=5 pid=45 OS id=2570 stopped  
Errors in file /u01/app/oracle/diag/rdbms/orcl2/orcl/trace/orcl_as01_2562.trc:  
ORA-00001: unique constraint (SYS.IDX_SCOTT_T_1) violated  
LOGMINER: session#=1 (Logical_Standby$), reader MS00 pid=34 OS id=2554 sid=34 stopped  
LOGMINER: session#=1 (Logical_Standby$), preparer MS02 pid=36 OS id=2558 sid=45 stopped  
LOGMINER: session#=1 (Logical_Standby$), builder MS01 pid=35 OS id=2556 sid=44 stopped





trace文件/u01/app/oracle/diag/rdbms/orcl2/orcl/trace/orcl_as02_2535.trc里有这个事务相关信息  

ERROR INFORMATION:  
PRIMARY TXN xid: 0x0003.003.0000033b (   XIDUSN = 3, XIDSLT = 3, XIDSQN = 827)  
start scn: 0x0000.0011024e (1114702) commit scn: 0x0000.00110250 (1114704)


视图里有可读性强的错误信息  

SQL> select event, status from DBA_LOGSTDBY_EVENTS;  

EVENT                            STATUS  
-------------------------------- -----------------------------------------------------------  
insert into "SCOTT"."T"          ORA-00001: unique constraint (SYS.IDX_SCOTT_T_1) violated  
values  
    "USERNAME" = 'SYS',  
    "USER_ID" = 0,  
    "CREATED" = '17-SEP-11'


。。。。。。。。。。。




根据上面的xidusn,xidslt,xidsqn跳过事务  

SQL> exec dbms_logstdby.skip_transaction (   3,    3,    827);  

PL/SQL procedure successfully completed.  


SQL> ALTER DATABASE START LOGICAL STANDBY APPLY IMMEDIATE;  

Database altered.  




或者直接从视图里拼出语句  

SQL> select event, status,   'exec dbms_logstdby.skip_transaction ('||XIDUSN||', '||XIDSLT||', '||XIDSQN||');' from dba_logstdby_events where XIDUSN is not null;  

EVENT              STATUS  
-------------------------------------------------- ----------------------------------------------------------------------------------------------------  
'EXECDBMS_LOGSTDBY.SKIP_TRANSACTION('||XIDUSN||','||XIDSLT||','||XIDSQN||');'  
-------------------------------------------------------------------------------------------------------------------------------------------------------------------  
insert into "SCOTT"."T"          ORA-00001: unique constraint (SYS.IDX_SCOTT_T_1) violated  
values  
    "USERNAME" = 'SYS',  
    "USER_ID" = 0,  
    "CREATED" = '17-SEP-11'  
exec dbms_logstdby.skip_transaction (3, 3, 827);  


。。。。。。。。。。。。。



还有一个办法,就是全同步单表  

1 在备库上创建到主库的dblink,要求dblink在主库那边的用户能够读表、锁表及SELECT_CATALOG_ROLE

create database link dg_orcl connect to system identified by oracle using 'dg_orcl';     

2 调用存储过程

exec dbms_logstdby.instantiate_table('SCOTT', 'T', 'DG_ORCL');


标签:11g,LOGMINER,备库,17,pid,故障,LOGSTDBY,OS,id
来源: https://blog.51cto.com/lhrbest/2715482