ORACLE RAC ADG备库报错ORA-04021: timeout occurred while waiting to lock object

问题：核心的灾备 RAC ADG 备库，这两天频繁重启，并且报如下错误，通过查看MOS，发现是个BUG

ADG备库的ALERT错误日志如下：
Errors in file /u01/app/oracle/diag/rdbms/hxxxsz/hxxxsz1/trace/hxxxsz1_lgwr_69711.trc:
ORA-04021: timeout occurred while waiting to lock object
Mon Dec 16 16:26:15 2024
ORA-01555 caused by SQL statement below (SQL ID: 87gaftwrm2h68, Query Duration=899 sec, SCN: 0x05cf.7a01a7dc):
select o.owner#,o.name,o.namespace,o.remoteowner,o.linkname,o.subname from obj$ o where o.obj#=:1
LGWR (ospid: 69711): terminating the instance due to error 4021
Mon Dec 16 16:26:15 2024
System state dump requested by (instance=1, osid=69711 (LGWR)), summary=[abnormal instance termination].
System State dumped to trace file /u01/app/oracle/diag/rdbms/hxxxsz/hxxxsz1/trace/hxxxsz1_diag_69557_20241216162615.trc
Mon Dec 16 16:26:15 2024
ORA-1092 : opitsk aborting process
Mon Dec 16 16:26:16 2024
License high water mark = 1321
Instance terminated by LGWR, pid = 69711
USER (ospid: 42412): terminating the instance
Instance terminated by USER, pid = 42412
Mon Dec 16 16:26:23 2024
Starting ORACLE instance (normal)
************************ Large Pages Information *******************
Per process system memlock (soft) limit = UNLIMITED

解决方案:

1. 查看隐藏参数：
SELECT ksppinm, ksppstvl, ksppdesc FROM   x$ksppi x, x$ksppcv y WHERE   x.indx = y.indx AND  ksppinm ='_adg_parselock_timeout';

KSPPINM
--------------------------------------------------------------------------------
KSPPSTVL
--------------------------------------------------------------------------------
KSPPDESC
--------------------------------------------------------------------------------
_adg_parselock_timeout
0
timeout for parselock get on ADG in centiseconds

2. 执行以下语句：
alter system set "_adg_parselock_timeout"=500 scope=both sid='*';

参考MOS内容如下：

ORA-04021: timeout occurred while waiting to lock object : DR Instance terminated by LGWR (Doc ID 2183882.1)

Applies to:

Oracle Database - Enterprise Edition - Version 11.2.0.3 and later Oracle Database Exadata Cloud Machine - Version N/A and later Oracle Cloud Infrastructure - Database Service - Version N/A and later Oracle Database Cloud Exadata Service - Version N/A and later Oracle Database Exadata Express Cloud Service - Version N/A and later Information in this document applies to any platform.

Symptoms

DR database crashed with below errors.. Client address: (ADDRESS=(PROTOCOL=<protocol>)(HOST=<hostname>)(PORT=<port>)) WARNING: inbound connection timed out (ORA-3136) Wed Jul 13 13:43:24 2016 Errors in file /<path>/diag/rdbms/<db_name>/<oracle_sid>/trace/<oracle_sid>_lgwr_<pid>.trc: ORA-04021: timeout occurred while waiting to lock object LGWR (ospid: 31312): terminating the instance due to error 4021 Wed Jul 13 13:43:24 2016 System state dump requested by (instance=1, osid=31312 (LGWR)), summary=[abnormal instance termination]. System State dumped to trace file /<path>/diag/rdbms/<db_name>/<oracle_sid>/trace/<oracle_sid>_diag_<pid>.trc Wed Jul 13 13:43:25 2016 License high water mark = 318 Instance terminated by LGWR, pid = 31312 USER (ospid: 20898): terminating the instance Instance terminated by USER, pid = 20898 Wed Jul 13 13:43:39 2016 Starting ORACLE instance (normal)

Cause

Bug 16717701 - ADG SHOULD GET THE INSTANCE PARSE LOCK WITH A TIMEOUT  ------> Superseded By Bug fix Bug 17018214 Bug 11712267 - ACTIVE DATA GUARD DATABASE HUNG ON 'LIBRARY CACHE: MUTEX X' WAIT EVENT LGWR trace file (RXEPRR1_lgwr_31312.trc) *** 2016-07-13 13:43:24.498 *** SESSION ID:(6709.1) 2016-07-13 13:43:24.498 *** CLIENT ID:() 2016-07-13 13:43:24.498 *** SERVICE NAME:(SYS$BACKGROUND) 2016-07-13 13:43:24.498 *** MODULE NAME:() 2016-07-13 13:43:24.498 *** ACTION NAME:() 2016-07-13 13:43:24.498 error 4021 detected in background process ORA-04021: timeout occurred while waiting to lock object kjzduptcctx: Notifying DIAG for crash event ----- Abridged Call Stack Trace ----- ksedsts()+1296<-kjzdicrshnfy()+364<-ksuitm()+1688<-ksbrdp()+4296<-opirip()+1680<-opidrv()+748<-sou2o()+88<-opimai_real()+276<-ssthrdmain()+316<-main()+316<-_start()+380 ----- End of Abridged Call Stack Trace -----

Solution

Issue matches with bug 11712267 and bug 16717701 Since two bugs are matching with the case, You can try with option (1) . As per Bug 11712267 change the cursor_sharing to force on Active dataguard (ADG). Monitor your environment for sometime. If it crashes again then follow with the option (2) Option (2): As per bug description LGWR can request DBINSTANCE lock in X mode without any timeout which can lead to a hang / deadlock. Both fixes are already included in 11.2.0.4 but the fix is DISABLED by default. == > To ENABLE the fix one has to set == > "_adg_parselock_timeout" > to the number of centi-seconds == > LGWR should wait before backing off and retrying the request. Value should be in centi seconds. == > I Don't think there is really any hard fast rule for a value - at default (0) it will not timeout. A value representing a few seconds seems reasonable - if LGWR has been stuck for say 5 seconds waiting it seems reasonable guess it is not going to get the lock. The param just causes it to abort the current attempt and retry If you want to play safe can start with a higher value then decrease later. A higher value will just mean more sessions blocked for longer in case of the deadlock situation. 500 Seems reasonable , but I have no data to base it on. There should be a statistic "ADG parselock X get attempts" If it gets set too small that value would likely increase a lot due to keep timing out and retrying. This is a dynamic parameter Follow option (1) . change the cursor_sharing to force on ADG If issue re-appears then follow option (2) as below Please set "_adg_parselock_timeout" to 500 == > SQL > alter system set "_adg_parselock_timeout"=500 scope=both sid='*';