Oracle DG主备架构中GAP归档裂缝处理方法总结一、概述当主库的某些日志没有成功传送到备库那么这时候就发生了归档裂缝Archive Gap。Oracle提供了两种日志GAP的检测和处理机制分别是自动GAP处理Automatic Gap Resolution和FAL进程GAP处理FAL Gap Resolution。**自动GAP处理**主库上的ARCn进程会每分钟检查备库上的日志GAP情况并做相应处理。**FALFetch Archive Log**通过配置FAL_SERVER和FAL_CLIENT实现GAP检测的一种机制**备库主动发起的“取”(fetch)**日志的过程。备库就是FAL_CLIENT它从FAL_SERVER中取这些GAP。特点1、Oracle会首先尝试使用FAL进程处理GAP当发现FAL机制并没有配置生效的时候进而尝试使用自动GAP处理。2、FAL进程只在物理备库存在。3、FAL进程提供了一个CLIENT/SERVER的机制用来解决检测在主库产生的连续的归档日志而备库接收的归档日志不连续的问题。4、FAL进程只有在需要的时候才会启动而当工作完成后就关闭了因此在正常情况下该进程是无法看见的。二、GAP处理方法当自动GAP处理和FAL进程GAP处理都无法处理GAP时就需要人工进行干预GAP处理方法主要有如下两类1、主库归档日志未丢失Oracle提供了两种日志GAP的检测和处理机制未正常处理主要有两种处理方法1通过11g版本开始的flush redo特性尝试处理2将主库的归档日志手工拷贝到备库服务器上并手工注册归档日志让备库识别和同步这些归档日志2、主库归档日志出现丢失这种情况需要人工干预来修复。三、GAP主库归档日志未丢失处理一flush redo方法在Oracle 11g开始Data Guard 切换多了一个新的功能flush redo。Flush 能把没有发送的redo 从主库传送到standby库。 只要主库能启动到mount 状态那么Flush 就可以把没有发送的归档和current online redo 发送到备库。Flush语法ALTER SYSTEM FLUSH REDO TO ‘target_db_name’ [ no confirm apply];注意1targe_db_name表示备库的db_unique_name2必须加上引号3主库Primary Database必须处于 Mount 状态才能执行该命令4no confirm apply 表示主库将redo数据发送到备库但不等待备库的应用确认就直接返回成功。此时命令变为异步操作只要主库确认 redo 数据已成功发送到备库的日志文件中命令就视为成功不再关心备库是否完成了应用。5不带no confirm apply表示主库将 redo 数据发送到备库并等待备库确认这些 redo 数据已经成功应用后才返回成功。此时FLUSH REDO 命令是一个同步操作它会阻塞直到备库确认所有发送过去的 redo 都已被成功应用二手工传输归档日志和注册的方法1、物理备库上执行如下SQL判断是否存在gapSELECT * FROM V$ARCHIVE_GAP; THREAD# LOW_SEQUENCE# HIGH_SEQUENCE# ----------- ------------- -------------- 1 40 432、主库上查询确认SELECT NAME FROM V$ARCHIVED_LOG WHERE THREAD#1 AND DEST_ID1 AND SEQUENCE# BETWEEN 41 AND 43; NAME ------------------------------------------------------------------------------- /oradata/arch/1_41_1236681669.dbf /oradata/arch/1_42_1236681669.dbf3、把这些归档日志复制、传输至物理备库服务器归档目录[oraclehost31 arch]$ scp 1_41_*.dbf 1_42_*.dbf host32:/oradata/arch oraclehost32s password: 1_41_1236681669.dbf 100% 58KB 49.6MB/s 00:00 1_42_1236681669.dbf 100% 20KB 25.7MB/s 00:00 备注如果需要的归档日志已经不在主库服务器上了但是有归档日志的RMAN备份那么可以通过RMAN恢复把缺少的归档日志进行还原类似如下 rman target / set archivelog destination to /oradata/arch; restore archivelog from logseq 40;4、备库上使用ALTER DATABASE REGISTER LOGFILE应用这些归档ALTER DATABASE REGISTER LOGFILE /oradata/arch/1_41_1236681669.dbf; ALTER DATABASE REGISTER LOGFILE /oradata/arch/1_42_1236681669.dbf; 备库上最大的日志序列号变为42表示已经自动应用手工注册的归档日志 select inst_id,max(sequence#) ,min(sequence#) from gv$archived_log group by inst_id order by 1; INST_ID MAX(SEQUENCE#) MIN(SEQUENCE#) ---------- -------------- -------------- 1 42 35备注注册完成后备库会自动应用这些归档日志完成主备同步5、可选如果后续是准备failover并打开数据库传输完主库的归档后可以继续使用主库的当前REDO恢复尽量确保数据恢复最大化甚至主备数据一致1主库上将redo日志传输至备库 scp /oradata/orcl/redo*.log host32:/oradata/orcl 2备库上确保备库mount状态 alter database recover managed standby database cancel; //先取消日志恢复 shutdown immediate; //关闭数据库 startup mount; //启动备库至mount状态 3备库执行不完全恢复 recover standby database until cancel; //备库执行不完全恢复通过主库传输过来的redo进行尝试 ORA-00279: change 1042670 generated at 06/29/2026 14:07:49 needed for thread 1 ORA-00289: suggestion : /oradata/arch/1_43_1236681669.dbf ORA-00280: change 1042670 for thread 1 is in sequence #43 Specify log: {RETsuggested | filename | AUTO | CANCEL} /oradata/orcl/redo01.log ----具体是哪些redo日志文件可以一一尝试 Log applied. Media recovery complete. 备注如果/oradata/orcl/redo01.log报不准确则继续尝试/oradata/orcl/redo02.log依次这样确保所有日志文件都完成恢复 4failover切换 alter database activate standby database; ---强制切换备库为主库 5打开备库 alter database open; select protection_mode,database_role,open_mode from v$database; PROTECTION_MODE DATABASE_ROLE OPEN_MODE -------------------- ---------------- -------------------- MAXIMUM PERFORMANCE PRIMARY READ WRITE四、GAP主库归档日志丢失处理(一)处理本质总结12c版本之前的常规处理步骤1、在主库上创建一个备库的控制文件2、以备库的当前SCN号为起点在主库上做一个增量备份3、将新创建的控制文件和增量备份拷贝到备库服务器上4、使用新的控制文件将备库启动到mount状态5、将增量备份注册到RMAN的catalog取消备库的恢复应用恢复增量备份6、开启备库的恢复进程上述六个步骤是处理归档日志丢失gap的本质过程12c版本、18c版本逐步可以做到一键式处理但是本质都是通过上述几个步骤完成只是封装后所需要执行的命令越来越简单12c开始处理命令RECOVER … FROM SERVICE18c开始处理命令RECOVER STANDBY DATABASE FROM SERVICE Oracle在18c对12c的功能进行了加强推出了(RECOVER STANDBY DATABASE … FROM SERVICE)通过网络服务直接增量恢复备库数据,这个新特性也大大简化了DataGuard主备的问题修复。Oracle随着版本的升级逐渐将步骤缩减进行封装18C之后可谓是达到了所谓的一键刷新恢复DG同步。二11g版本归档日志丢失gap处理案例gap模拟说明模拟备库断电主库做一些产生日志的动作然后切几个最新的归档最后手工删掉这几个归档日志备库重新开启DG同步。1、模拟归档日志丢失gap1备库down掉模拟断电 alter database recover managed standby database cancel; shutdown immediate 2主库做一些产生日志的动作然后多切换几个归档日志 insert into t1 values(10); commit; alter system archive log current; create table tab_user1 as select * from dba_users; alter system archive log current; insert into t1 values(11); alter system archive log current; 3主库服务器上删除最近几个归档日志 cd /oradata/arch mkdir bak mv 1_47_1236681669.dbf 1_48_1236681669.dbf 1_49_1236681669.dbf bak/ 4启动备库开启同步 startup ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE DISCONNECT FROM SESSION; 5备库上查看GAP情况 select * from v$archive_gap; THREAD# LOW_SEQUENCE# HIGH_SEQUENCE# ---------- ------------- -------------- 1 46 49 SELECT max(sequence#) from v$archived_log where appliedYES; MAX(SEQUENCE#) -------------- 46 set pagesize 99 SELECT PROCESS, STATUS, THREAD#, SEQUENCE#, BLOCK#, BLOCKS FROM V$MANAGED_STANDBY; PROCESS STATUS THREAD# SEQUENCE# BLOCK# BLOCKS --------- ------------ ---------- ---------- ---------- ---------- ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CLOSING 1 50 1 114 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 RFS IDLE 0 0 0 0 RFS IDLE 0 0 0 0 RFS IDLE 1 51 2005 1 MRP0 APPLYING_LOG 1 47 65 102400 12 rows selected.最终GAP归档日志确认为47—49这三个归档日志2、正式处理过程1在主库上创建一个备库的控制文件alter database create standby controlfile as /oradata/backup/standby.ctl; Database altered.2以备库的当前SCN号为起点在主库上做一个增量备份备库上执行查询备库的当前scn号 SQL select to_char(current_scn) from v$database; TO_CHAR(CURRENT_SCN) ---------------------------------------- 1044264 主上执行你以备库scn号为起点主库做一个增量备份 rman target / backup INCREMENTAL from scn 1044264 database format /oradata/backup/incre_%U; 备注可以并发方式完成类似如下 run{ allocate channel c1 type disk; allocate channel c2 type disk; backup INCREMENTAL from scn 1086639 database format /oradata/backup/incre_%U; release channel c1; release channel c2; }3将控制文件和增量备份拷贝到备库上主库服务器上执行 scp /oradata/backup/* oraclehost32:/oradata/backup/ [oraclehost31 backup]$ scp /oradata/backup/* oraclehost32:/oradata/backup/ oraclehost32s password: incre_144rsqmg_1_1 100% 3032KB 103.6MB/s 00:00 incre_154rsqmh_1_1 100% 9856KB 82.4MB/s 00:00 standby.ctl 100% 9808KB 212.2MB/s 00:004使用新的控制文件将备库启动到mount状态shutdown immediate startup nomount rman target / restore controlfile from /oradata/backup/standby.ctl; 具体输出 Starting restore at 29-JUN-26 using target database control file instead of recovery catalog allocated channel: ORA_DISK_1 channel ORA_DISK_1: SID20 device typeDISK channel ORA_DISK_1: copied control file copy output file name/oradata/orcl/control01.ctl output file name/oradata/orcl/control02.ctl Finished restore at 29-JUN-26 启动至mount状态 alter database mount;5增量备份注册到RMAN的catalog取消日志应用恢复增量备份备库取消日志应用 ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL; 增量备份注册到备库的catalog中 rman target / catalog start with /oradata/backup; 具体输出 RMAN catalog start with /oradata/backup; released channel: ORA_DISK_1 searching for all files that match the pattern /oradata/backup List of Files Unknown to the Database File Name: /oradata/backup_bak/st1.ctl File Name: /oradata/backup/incre_144rsqmg_1_1 File Name: /oradata/backup/incre_154rsqmh_1_1 File Name: /oradata/backup/standby.ctl Do you really want to catalog the above files (enter YES or NO)? YES //输入YES cataloging files... cataloging done List of Cataloged Files File Name: /oradata/backup_bak/st1.ctl File Name: /oradata/backup/incre_144rsqmg_1_1 File Name: /oradata/backup/incre_154rsqmh_1_1 File Name: /oradata/backup/standby.ctl 恢复增量备份 recover database noredo; 具体输出 Starting recover at 29-JUN-26 allocated channel: ORA_DISK_1 channel ORA_DISK_1: SID20 device typeDISK channel ORA_DISK_1: starting incremental datafile backup set restore channel ORA_DISK_1: specifying datafile(s) to restore from backup set destination for restore of datafile 00001: /oradata/orcl/system01.dbf destination for restore of datafile 00002: /oradata/orcl/sysaux01.dbf destination for restore of datafile 00003: /oradata/orcl/undotbs01.dbf destination for restore of datafile 00004: /oradata/orcl/users01.dbf channel ORA_DISK_1: reading from backup piece /oradata/backup/incre_144rsqmg_1_1 channel ORA_DISK_1: piece handle/oradata/backup/incre_144rsqmg_1_1 tagTAG20260629T150552 channel ORA_DISK_1: restored backup piece 1 channel ORA_DISK_1: restore complete, elapsed time: 00:00:01 Finished recover at 29-JUN-266开启备库的恢复进程备库只读方式打开并且启动应用 alter database open read only; ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE DISCONNECT FROM SESSION; 查询视图确认GAP已修复 SELECT * FROM V$ARCHIVE_GAP; SELECT max(sequence#) from v$archived_log where appliedYES; SELECT PROCESS, STATUS, THREAD#, SEQUENCE#, BLOCK#, BLOCKS FROM V$MANAGED_STANDBY; PROCESS STATUS THREAD# SEQUENCE# BLOCK# BLOCKS --------- ------------ ---------- ---------- ---------- ---------- ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CLOSING 1 51 2048 1064 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 ARCH CONNECTED 0 0 0 0 RFS IDLE 0 0 0 0 RFS IDLE 0 0 0 0 RFS IDLE 1 52 143 1 MRP0 APPLYING_LOG 1 52 143 102400 12 rows selected. SQL select * from t1; ID ---------- 10 11至此完成11g版本归档日志丢失的GAP修复三12c版本归档日志丢失gap处理案例采用rman的新功能recover standby using service通过RMAN连接到target备库然后用主库的service执行恢复备库命令。语法 RECOVER DATABASE FROM SERVICE PRIMARY DB SERVICE NAME NOREDO USING COMPRESSED BACKUPSET;具体恢复过程备库启动到nomount状态 shutdown immediate startup nomount rman恢复 rman target / 恢复控制文件 restore standby controlfile from service orcl; 备库mount alter database mount; 执行恢复 recover database from service orcl noredo using compressed backupset; 备注如果期间有新增数据文件目录位置不一样需要进行额外的处理工作 开启备库应用日志 alter database open; alter pluggable database all open; ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE DISCONNECT FROM SESSION;四18c版本及以上归档日志丢失gap处理案例采用RECOVER STANDBY DATABASE FROM SERVICE恢复备库RECOVER STANDBY DATABASE命令完成的工作1重新启动备用实例;2)从主数据库刷新控制文件;3)并自动重命名数据文件临时文件和联机日志;4)它可以还原添加到主数据库中的新数据文件;5)还原到当前时间的备用数据库。具体恢复过程备库取消日志应用 ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL; 开始恢复备库备库上执行 rman target / RECOVER STANDBY DATABASE FROM SERVICE orcl; 开启备库应用日志 alter database open; alter pluggable database all open; ALTER DATABASE RECOVER MANAGED STANDBY DATABASE USING CURRENT LOGFILE DISCONNECT FROM SESSION;