oracle報錯ORA-00353: log corruption due to GoldenGate extract process問題

某客戶今天告訴我,一套核心庫的備庫的日誌漲的很快，通過檢查發現alert log不斷的報錯，如下所示：

代碼如下

複製代碼

Fri Dec 11 20:48:37 2015 Incomplete read from log member '/arch2/2_82548_864474703.dbf'. Trying next member. Incomplete read from log member '/arch2/2_82548_864474703.dbf'. Trying next member.Archived Log entry 101817 added for thread 2 sequence 82548 ID 0xedd14d2a dest 1: Incomplete read from log member '/arch2/2_82548_864474703.dbf'. Trying next member. Errors in file /oracle/app/oracle/diag/rdbms/crmadg/crmdb2/trace/crmdb2_ora_15304.trc  (incident=711224): ORA-00353: log corruption near block 900532 change 14410909668198 time 12/11/2015 20:42:40 ORA-00334: archived log: '/arch2/2_82548_864474703.dbf' Errors in file /oracle/app/oracle/diag/rdbms/crmadg/crmdb2/trace/crmdb2_ora_15305.trc  (incident=711216): ORA-00353: log corruption near block 900532 change 14410909668198 time 12/11/2015 20:42:40 ORA-00334: archived log: '/arch2/2_82548_864474703.dbf' Fri Dec 11 20:48:37 2015 Errors in file /oracle/app/oracle/diag/rdbms/crmadg/crmdb2/trace/crmdb2_ora_13484.trc  (incident=711238): ORA-00353: log corruption near block 900532 change 14410909668198 time 12/11/2015 20:42:40 ORA-00334: archived log: '/arch2/2_82548_864474703.dbf' Fri Dec 11 20:49:13 2015 Sweep [inc][711238]: completed

查看其中的trace可以看到如下的內容：

代碼如下	複製代碼
*** ACTION NAME:() 2015-12-10 21:58:10.036   kcrfrxini: Limitread is not enabled for standby db kcrfrxini: Limitread is not enabled for standby db kcrfrxini: Limitread is not enabled for standby db

....
我google一把，發現居然毫無所獲。。。從alert log最開始報錯的時間點開始看，我懷疑跟最近部署的goldengate有關係。
後面我搜尋索引鍵ORA-00353 log corruption  ogg extract 發現mos 論壇上有人遇到類似的問題，發帖者提到使用了如下的參數：

代碼如下	複製代碼
TranLogOptions DBLOGREADER TranLogOptions BUFSIZE 4096000 TranLogOptions DBLOGREADERBUFSIZE 4096000

該兄弟的環境為oracle 11.2.0.3 rac，ogg 版本為11.2.0.1.5；
最後據說通過將參數buffersize 、DBLOGREADERBUFSIZE 調小來解決該問題，最後貌似並沒有提到解決了，只是暫時未出現。

考慮到客戶這裡是Oracle 11.2.0.4環境，雖然是ASM環境，然而歸檔日誌都存放到本地檔案系統上。
再次檢查goldengate extract參數：

代碼如下

複製代碼

ogg_eda@xxx:/ogg_eda$ ./ggsci   Oracle GoldenGate Command Interpreter for Oracle Version 11.2.1.0.27 19591627 OGGCORE_11.2.1.0.0OGGBP_PLATFORMS_141006.1156 HP/UX, IA64, 64bit (optimized), Oracle 11g on Oct  6 2014 13:37:58   Copyright (C) 1995, 2014, Oracle and/or its affiliates. All rights reserved.   GGSCI (xxx) 2> info all   Program     Status      Group       Lag at Chkpt  Time Since Chkpt   MANAGER     RUNNING EXTRACT     RUNNING     DPEODS1     00:00:00      00:00:00 EXTRACT     RUNNING     DPEODS2     00:00:00      00:00:00 EXTRACT     RUNNING     DPEODS3     00:00:00      00:00:07 EXTRACT     RUNNING     DPEODS4     00:00:00      00:00:08 EXTRACT     RUNNING     EXTODS1     00:00:00      00:00:06 EXTRACT     RUNNING     EXTODS2     00:00:00      00:00:00 EXTRACT     RUNNING     EXTODS3     00:00:00      00:00:03 EXTRACT     RUNNING     EXTODS4     00:00:00      00:00:03      GGSCI (xxx) 3> view param EXTODS1   EXTRACT extods1 setenv (NLS_LANG = "AMERICAN_AMERICA.ZHS16GBK") SETENV (ORACLE_HOME = "/oracle/app/oracle/product/11.2.0") SETENV (ORACLE_SID = "xxx2") USERID ogg_eda, PASSWORD AACAAAAAAAAAAAIARCQIHGOIWCQJXITB ,ENCRYPTKEY default TRANLOGOPTIONS ALTARCHIVELOGDEST primary instance crmdb2 /arch2,ALTARCHIVELOGDEST primary instance crmdb1 /arch2 TRANLOGOPTIONS ARCHIVEDLOGONLY TRANLOGOPTIONS ALTARCHIVEDLOGFORMAT %t_%s_%r.dbf TranlogOptions DBLOGREADER REPORTCOUNT EVERY 1 MINUTES, RATE DISCARDFILE ./dirrpt/extods1.dsc,APPEND,MEGABYTES 1024 DBOPTIONS  ALLOWUNUSEDCOLUMN WARNLONGTRANS 2h,CHECKINTERVAL 3m EXTTRAIL ./dirdat/ods/o1 FETCHOPTIONS NOUSESNAPSHOT 。。。。。

由於是抽取active dataguard且僅僅抽取archivelog，因此完全沒有必要加入TranlogOptions DBLOGREADER參數。
這裡我懷疑跟該參數有關，通過屏蔽該參數後，發現不在報錯。看來這是GoldengGate的未知bug呀！