Legato备份还原测试

· 作者 小荷||【转载时请务必以超链接形式标明文章原始出处和作者信息】
· 永久链接: http://www.oracleblog.cn/working-case/test-legato-recover/

最近的一段时间，某省的备份出了点问题，legato的备份在还原时检查不到索引，通过scanner扫描之后还是检查不到，问题已经开case提交到emc那边，目前还在处理。为了防止万一，还是对全网用legato备份的省进行了一次还原测试。以下是还原测试中遇到的一些问题小结。

目前现网大部分机器都是采用hp service guard，primary机上跑着生成，standby机器上已经装好了oracle软件，但是是处于standby状态。还原的内容是在备机上将spfile还原出来，大致的步骤如下：

1、在legato上的group中反击右键，在save set中看到备份脚本的路径，一般是在$ORACLE_HOME/bin下。
2、去$ORACLE_HOME/dbs下把原来的pfile做一下mv
3、rman target / catalog rman/rman@xxrman 这个登录的密码根据第1步中找到的登录方式，进行登录
4、做startup nomount
5、指定incarnation
6、还原spfile，把之前第1步的脚本改改，只分配一个channel就可以，把backup部分去掉，换成restore spfile。在这里有一些注意点，在后面会讲到。
7、检查和扫尾。
7.1、检查在dbs下是否有spfile的生成。
7.2.1、把恢复出来的spfile干掉
7.2.2、把initgzmisc.ora.bak 再mv回去。
7.2.3、停测试用的数据库

问题1：在备份脚本中虽然只写了NSR_CLIENT，没写NSR_SERVER，但是在还原测试的时候，需要写上NSR_SERVER。不然会报错：

RMAN> run {
2>   allocate channel t1 type 'sbt_tape'
  parms 'ENV=(NSR_CLIENT=gx_db)';
  restore spfile;
  release channel t1;
}3> 4> 5> 6>
 
allocated channel: t1
channel t1: sid=9 devtype=SBT_TAPE
channel t1: NMO v4.1.0.0
 
Starting restore at 30-DEC-09
 
channel t1: starting datafile backupset restore
channel t1: restoring SPFILE
output filename=/oracle/app/oracle/product/9.2.0/dbs/spfilegxmisc.ora
released channel: t1
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of restore command at 12/30/2009 14:04:00
ORA-19507: failed to retrieve sequential file, handle="db_GXMISC_t706434758_s43456_p1", parms=""
ORA-27029: skgfrtrv: sbtrestore returned error
ORA-19511: Error received from media manager layer, error text:
   nwora_open_restore: Could not locate the NWORA save file 'db_GXMISC_t706434758_
s43456_p1' on server 'GX_BAK01'.

问题2：在主机2上做还原，但是脚本中的NSR_CLIENT还是得写备份时候的，如备份脚本中用的是xx_db01，那么在xx_db02上还原的时候，不能写NSR_CLIENT=xx_db02，应该写xx_db01.不然会报错：

RMAN> run {
2>  allocate channel t1 type 'sbt_tape'
parms 'ENV=(NSR_CLIENT=sn_dc02)';
restore spfile;
release channel t1;
} 3> 4> 5> 6>
 
allocated channel: t1
channel t1: sid=9 devtype=SBT_TAPE
channel t1: NMO v4.1.0.0
 
Starting restore at 30-DEC-09
 
channel t1: starting datafile backupset restore
channel t1: restoring SPFILE
output filename=/oracle/app/oracle/product/9.2.0s/spfilesnmisc.ora
released channel: t1
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of restore command at 12/30/2009 15:10:31
ORA-19507: failed to retrieve sequential file, handle="db_SNMISC_t706845618_s19848_p1", parms=""
ORA-27029: skgfrtrv: sbtrestore returned error
ORA-19511: Error received from media manager layer, error text:
   nwora_open_restore: Could not locate the NWORA save file 'db_SNMISC_t706845618_s19848_p1' on server 'sn-bak01'.

问题3：如果主机开启了安全加固，不允许rlogin，rexec之类的操作，且在xx_db01的脚本中配置的NSR_CLIENT不是浮动IP，那么在还原的时候也会报错。此时需要修改备份脚本，将NSR_CLIENT指定为浮动IP的hostname。

RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of restore command at 12/31/2009 11:39:02
ORA-19507: failed to retrieve sequential file, handle="db_GXMISC_t707018793_s43636_p1", parms=""
ORA-27029: skgfrtrv: sbtrestore returned error
ORA-19511: Error received from media manager layer, error text:
   nwora_open_restore: The NW authentication for client 'gx_db01' was refused by server 'gx_bak01' because 'User oracle on computer gx_db02 is not on gx_db01's remote access list'.
RMAN> exit

问题4：如果还原的时候，在client端的hosts没配置好也会报错：
在rman的脚本中会提示：

RMAN> run {
2>  allocate channel t1 type 'sbt_tape'
 parms 'ENV=(NSR_CLIENT=yn_db,NSR_SERVER=yn_bak)';
 restore spfile;
 release channel t1;
 } 3> 4> 5> 6>
allocated channel: t1
channel t1: sid=10 devtype=SBT_TAPE
channel t1: NMO v4.1.0.0
Starting restore at 05-JAN-10
channel t1: starting datafile backupset restore
channel t1: restoring SPFILE
output filename=/oracle/app/oracle/product/9.2.0/dbs/spfileynmisc.ora
okreleased channel: t1
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of restore command at 01/05/2010 11:26:54
ORA-19507: failed to retrieve sequential file, handle="db_YNMISC_t707467406_s16566_p1", parms=""
ORA-27029: skgfrtrv: sbtrestore returned error
ORA-19511: Error received from media manager layer, error text:
   nwora_open_restore: Could not get mmd connection.
RMAN>

此时，在legato的界面中看到：
legato中显示的log有yn_sdb
此时我们看到还原的时候，用了一个rd=yn_sdb的设备，我们在server从查到yn_sdb是一个审计数据库，而不是我们当前的生产库。

同时我们也在client端看到legato的log：

oracle@yn_dc02:/nsr/applogs>ll
total 16
-rw-r--r--   1 oracle     dba            849 Jan  5 11:26 nmo.messages
oracle@yn_dc02:/nsr/applogs>more *
SBT-7878 12/30/09 15:51:57 nwora_restore_mmdconn: could not set up mm binding: Unknown host
SBT-7878 12/30/09 15:51:57 nwora_open_restore: Could not get mmd connection.
SBT-3980 12/31/09 10:57:03 nwora_restore_mmdconn: could not set up mm binding: Unknown host
SBT-3980 12/31/09 10:57:03 nwora_open_restore: Could not get mmd connection.
SBT-5655 12/31/09 12:10:38 nwora_restore_mmdconn: could not set up mm binding: Unknown host
SBT-5655 12/31/09 12:10:38 nwora_open_restore: Could not get mmd connection.
SBT-12983 12/31/09 17:26:24 nwora_restore_mmdconn: could not set up mm binding: Unknown host
SBT-12983 12/31/09 17:26:24 nwora_open_restore: Could not get mmd connection.
SBT-18279 01/05/10 11:26:54 nwora_restore_mmdconn: could not set up mm binding: Unknown host
SBT-18279 01/05/10 11:26:54 nwora_open_restore: Could not get mmd connection.
oracle@yn_dc02:/nsr/applogs>

有一个unknown的host，我们检查了该client端的hosts文件，发现确实没有yn_sdb的解析。

我们在做恢复的client端的hosts配上yn_sdb，再次测试还原，成功。

另外，检查了该server端的配置，发现remote driver的配置有误，dc01的第一个最后少了个字母b：

重新jbconfig之后，再次测试还原控制文件，发现没有在使用yn_sdb了：

· 【文章发布信息】发表于： 2010-01-04 @ 15:20:49 · ||分类： ..experience, Working case

留条评论