IT博客汇 | ORA-15042: ASM disk “N” is missing from group number “M” 故障恢复

ORA-15042: ASM disk “N” is missing from group number “M” 故障恢复

惜分飞发表于 2015-10-08 02:28:23

标题：ORA-15042: ASM disk “N” is missing from group number “M” 故障恢复

接到一个朋友恢复请求,19个lun的asm 磁盘组，由于其中一个lun有问题，他们进行了增加一个新lun，删除老lun的方法操作,但是操作一半hang住了(因为坏的lun是底层损坏，无法完成rebalance),然后存储工程师继续修复异常lun，非常幸运异常lun修复好了，但是高兴过了头，直接从存储上删除了新加入的lun(已经rebalance一部分数据进去了)，这个时候asm dg彻底趴下了，不能mount成功，请求恢复支持。由于某种原因，无法从lun层面恢复，只能让我们提供数据库层面恢复

Mon Sep 21 19:52:35 2015
SQL> alter diskgroup  dg_XFF add disk '/dev/rhdisk116' size 716800M drop disk dg_XFF_0012 
NOTE: Assigning number (1,20) to disk (/dev/rhdisk116)
NOTE: requesting all-instance membership refresh for group=1
NOTE: initializing header on grp 1 disk DG_XFF_0020
NOTE: requesting all-instance disk validation for group=1
Mon Sep 21 19:52:44 2015
NOTE: skipping rediscovery for group 1/0xb94738f1 (DG_XFF) on local instance.
NOTE: requesting all-instance disk validation for group=1
NOTE: skipping rediscovery for group 1/0xb94738f1 (DG_XFF) on local instance.
NOTE: initiating PST update: grp = 1
Mon Sep 21 19:52:44 2015
GMON updating group 1 at 25 for pid 27, osid 12124486
NOTE: PST update grp = 1 completed successfully 
NOTE: membership refresh pending for group 1/0xb94738f1 (DG_XFF)
GMON querying group 1 at 26 for pid 18, osid 10092734
NOTE: cache opening disk 20 of grp 1: DG_XFF_0020 path:/dev/rhdisk116
GMON querying group 1 at 27 for pid 18, osid 10092734
SUCCESS: refreshed membership for 1/0xb94738f1 (DG_XFF)
Mon Sep 21 19:52:47 2015
SUCCESS: alter diskgroup  dg_XFF add disk '/dev/rhdisk116' size 716800M drop disk dg_XFF_0012
NOTE: starting rebalance of group 1/0xb94738f1 (DG_XFF) at power 1
Starting background process ARB0
Mon Sep 21 19:52:47 2015
ARB0 started with pid=28, OS id=10944804 
NOTE: assigning ARB0 to group 1/0xb94738f1 (DG_XFF) with 1 parallel I/O
NOTE: Attempting voting file refresh on diskgroup DG_XFF
Mon Sep 21 20:35:06 2015

SQL> ALTER DISKGROUP DG_XFF MOUNT  /* asm agent *//* {1:51107:7083} */ 
NOTE: cache registered group DG_XFF number=1 incarn=0xdd6f975a
NOTE: cache began mount (first) of group DG_XFF number=1 incarn=0xdd6f975a
NOTE: Assigning number (1,0) to disk (/dev/rhdisk10)
NOTE: Assigning number (1,1) to disk (/dev/rhdisk11)
NOTE: Assigning number (1,2) to disk (/dev/rhdisk16)
NOTE: Assigning number (1,3) to disk (/dev/rhdisk17)
NOTE: Assigning number (1,4) to disk (/dev/rhdisk22)
NOTE: Assigning number (1,5) to disk (/dev/rhdisk23)
NOTE: Assigning number (1,6) to disk (/dev/rhdisk28)
NOTE: Assigning number (1,7) to disk (/dev/rhdisk29)
NOTE: Assigning number (1,8) to disk (/dev/rhdisk33)
NOTE: Assigning number (1,9) to disk (/dev/rhdisk34)
NOTE: Assigning number (1,10) to disk (/dev/rhdisk4)
NOTE: Assigning number (1,11) to disk (/dev/rhdisk40)
NOTE: Assigning number (1,12) to disk (/dev/rhdisk41)
NOTE: Assigning number (1,13) to disk (/dev/rhdisk45)
NOTE: Assigning number (1,14) to disk (/dev/rhdisk46)
NOTE: Assigning number (1,15) to disk (/dev/rhdisk5)
NOTE: Assigning number (1,16) to disk (/dev/rhdisk52)
NOTE: Assigning number (1,17) to disk (/dev/rhdisk53)
NOTE: Assigning number (1,18) to disk (/dev/rhdisk57)
NOTE: Assigning number (1,19) to disk (/dev/rhdisk58)
Wed Sep 30 11:08:07 2015
NOTE: start heartbeating (grp 1)
GMON querying group 1 at 33 for pid 35, osid 4194488
NOTE: Assigning number (1,20) to disk ()
GMON querying group 1 at 34 for pid 35, osid 4194488
NOTE: cache dismounting (clean) group 1/0xDD6F975A (DG_XFF) 
NOTE: dbwr not being msg'd to dismount
NOTE: lgwr not being msg'd to dismount
NOTE: cache dismounted group 1/0xDD6F975A (DG_XFF) 
NOTE: cache ending mount (fail) of group DG_XFF number=1 incarn=0xdd6f975a
NOTE: cache deleting context for group DG_XFF 1/0xdd6f975a
GMON dismounting group 1 at 35 for pid 35, osid 4194488
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
NOTE: Disk  in mode 0x8 marked for de-assignment
ERROR: diskgroup DG_XFF was not mounted
ORA-15032: not all alterations performed
ORA-15040: diskgroup is incomplete
ORA-15042: ASM disk "20" is missing from group number "1" 
ERROR: ALTER DISKGROUP DG_XFF MOUNT  /* asm agent *//* {1:51107:7083} */

这里比较明显，由于存储工程师直接删除了lun，这里导致磁盘组DG_XFF丢失asm disk 20，使得磁盘组无法直接mount，由于该磁盘组已经进行了较长时间的rebalance，丢失的盘中已经有大量数据（包括元数据），因此就算修改pst让磁盘组mount起来(不一定成功)，也会丢失大量数据，也不一定可以直接拿出来里面的数据，如果只是加入盘，但是由于某种原因没有做rebalance，那我们直接可以通过修改pst,使得磁盘组mount起来。因此对于这样的情况，我们能够做的，只能从底层扫描磁盘，生成数据文件(因为有部分文件的元数据在丢失lun之上，如果直接使用现存元数据信息，直接拷贝，或者unload数据都会丢失大量数据），然后再进一步unload数据，完成恢复。需要恢复磁盘信息

grp# dsk# bsize ausize disksize diskname        groupname       path
---- ---- ----- ------ -------- --------------- --------------- -------------
   1    0  4096  4096K   179200 DG_XFF_0000     DG_XFF          /dev/rhdisk10
   1    1  4096  4096K   179200 DG_XFF_0001     DG_XFF          /dev/rhdisk11
   1    2  4096  4096K   179200 DG_XFF_0002     DG_XFF          /dev/rhdisk16
   1    3  4096  4096K   179200 DG_XFF_0003     DG_XFF          /dev/rhdisk17
   1    4  4096  4096K   179200 DG_XFF_0004     DG_XFF          /dev/rhdisk22
   1    5  4096  4096K   179200 DG_XFF_0005     DG_XFF          /dev/rhdisk23
   1    6  4096  4096K   179200 DG_XFF_0006     DG_XFF          /dev/rhdisk28
   1    7  4096  4096K   179200 DG_XFF_0007     DG_XFF          /dev/rhdisk29
   1    8  4096  4096K   179200 DG_XFF_0008     DG_XFF          /dev/rhdisk33
   1    9  4096  4096K   179200 DG_XFF_0009     DG_XFF          /dev/rhdisk34
   1   10  4096  4096K   179200 DG_XFF_0010     DG_XFF          /dev/rhdisk4
   1   11  4096  4096K   179200 DG_XFF_0011     DG_XFF          /dev/rhdisk40
   1   12  4096  4096K   179200 DG_XFF_0012     DG_XFF          /dev/rhdisk41
   1   13  4096  4096K   179200 DG_XFF_0013     DG_XFF          /dev/rhdisk45
   1   14  4096  4096K   179200 DG_XFF_0014     DG_XFF          /dev/rhdisk46
   1   15  4096  4096K   179200 DG_XFF_0015     DG_XFF          /dev/rhdisk5
   1   16  4096  4096K   179200 DG_XFF_0016     DG_XFF          /dev/rhdisk52
   1   17  4096  4096K   179200 DG_XFF_0017     DG_XFF          /dev/rhdisk53
   1   18  4096  4096K   179200 DG_XFF_0018     DG_XFF          /dev/rhdisk57
   1   19  4096  4096K   179200 DG_XFF_0019     DG_XFF          /dev/rhdisk58

这次运气比较好，丢失的磁盘组只是一个业务磁盘组，而且里面只有19个表空间，10个分区表，因此在数据字典完成的情况下，恢复10个分区表(一共6443个分区)的数据，整体恢复效果如下:
RECOVER

从整体数据量看恢复比例为：6003.26953/6027.26935*100%=99.6018127%,对于丢失了一个已经rebalance的大部分的lun，依旧能够恢复如此的数据，整体看非常理想.