ADG terminated by LGWR, terminating the instance

11.2.0.4 RAC TO RAC FOR ADG環境。由于歷史原因，備庫節點二一直沒有啟動，一直是啟動節點一對外提供服務。節點一alert報錯，lgwr進行kill實例操作并自行重啟。
Mon Dec 24 16:11:24 2018
Archived Log entry 262740 added for thread 2 sequence 185858 ID 0x92570693 dest 1:
Mon Dec 24 16:12:28 2018
Exception [type: SIGSEGV, Address not mapped to object] [ADDR:0x7FFCD7489FF8] [PC:0x9899B16, qcsAnalyzeBooleanExpr()+144] [flags: 0x0, count: 1]
Errors in file /u01/app/oracle/diag/rdbms/qdb/qdb1/trace/qdb1_ora_18912.trc (incident=580470):
ORA-07445: exception encountered: core dump [qcsAnalyzeBooleanExpr()+144] [SIGSEGV] [ADDR:0x7FFCD7489FF8] [PC:0x9899B16] [Address not mapped to object] []
Incident details in: /u01/app/oracle/diag/rdbms/qdb/qdb1/incident/incdir_580470/qdb1_ora_18912_i580470.trc
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.

根據拋出的trc文件，我們可以追蹤到時間段內的某條SQL

** 2018-12-24 16:12:28.596
dbkedDefDump(): Starting a non-incident diagnostic dump (flags=0x3, level=3, mask=0x0)
----- Current SQL Statement for this session (sql_id=34vxxxxxx) -----
select from MSS.T_MSS_MOBILE_LOGIN_ADDR where UUID in ('7xxxxx4xxxxxx', '48xxxxxxx, 。。。。。。
整個trc文件190M，文本內容UUID幾乎占了整個文件。

new 2: where owner = 'MSS' and segment_name = 'T_MSS_MOBILE_LOGIN_ADDR'

OWNER SEGMENT_NAME SEGMENT_TYPE Total_Bytes(MB)

MSS T_MSS_MOBILE_LOGIN_ADDR TABLE PARTITION 230054

[oracle@qdb1 trace]$ du -sh /u01/app/oracle/diag/rdbms/qdb/qdb1/incident/incdir_580470/qdb1_ora_18912_i580470.trc
190M /u01/app/oracle/diag/rdbms/qdb/qdb1/incident/incdir_580470/qdb1_ora_18912_i580470.trc

緊接著系統報了4021錯誤代碼
Mon Dec 24 16:30:17 2018
System State dumped to trace file /u01/app/oracle/diag/rdbms/qdb/qdb1/trace/qdb1_ora_67693.trc
Mon Dec 24 16:32:32 2018
Errors in file /u01/app/oracle/diag/rdbms/qdb/qdb1/trace/qdb1_lgwr_39157.trc:
ORA-04021: timeout occurred while waiting to lock object
LGWR (ospid: 39157): terminating the instance due to error 4021
一直到最后的kill 實例，自動重啟
Mon Dec 24 16:32:39 2018
License high water mark = 1203
Instance terminated by LGWR, pid = 39157
USER (ospid: 6702): terminating the instance
Instance terminated by USER, pid = 6702
Mon Dec 24 16:32:48 2018
Starting ORACLE instance (normal)

查詢MOS相關文檔，有一篇文檔與我們的環境相符
ORA-04021: timeout occurred while waiting to lock object : DR Instance terminated by LGWR (文檔 ID 2183882.1)

命中了BUG了。根據bug描述，需要修改參數
SQL> show parameter cursor_sharing

NAME TYPE VALUE

cursor_sharing string EXACT
在cursor設置為exact時，兩條sql語句如果存在一點不同，就不會共享cursor，而進行兩次硬解析。
設置為force時Oracle對輸入的SQL值，會將where條件取值自動替換為綁定變量。以后在輸入相同的結構SQL語句時，會進行cursor sharing共享游標
根據這個臨時方案，我做了一個小實驗。我在主庫創建了一個t表作為該參數的測試

14:35:02 SYS@bapdb1(bapdb1)> create table t as select * from dba_objects;

Table created.

然后到備庫進行具體實驗。

Mark一下，擇日進行參數修改。