#ASM 翻译系列第三十二弹:ASM INTERNAL Find block in ASM

原文: Find block in ASM
作者: Bane Radulovic
译者:魏兴华,沃趣科技高级技术专家,主要参与公司一体机产品、监控产品、容灾产品、DBaaS平台的研发和设计。曾就职于东软集团,阿里巴巴集团,Oracle ACE组成员,DBGEEK 用户组发起人,ITPUB认证博客专家,ACOUG、SHOUG核心成员。曾在中国数据库大会、Oracle技术嘉年华、ORCL-CON、YY分享平台等公开场合多次做过数据库技术专题分享。对Oracle 并行机制、数据库异常恢复方法、ASM等有深入的研究,人称”Oracle Internal达人”,对企业数据库架构设计、故障恢复、高并发下数据库性能调优有丰富的经验,擅长从等待事件角度分析解决数据库性能问题,是OWI方法论的提倡者和践行者。
审校:魏兴华
责编:仲培艺

在本系列文章《Where is my data》中,我已经演示了如何从ASM磁盘中定位和抽取一个Oracle的block,为了让这件事做起来不那么复杂,我又写了一个perl脚本find_block.pl来简化整个操作,只需要提供数据文件的名称和需要提取的block,这个脚本就可以输出从ASM磁盘组中抽取块的命令。

find_block.pl

find_block.pl是一个perl脚本,脚本里集成了dd或kfed命令来从ASM磁盘中抽取一个块,脚本可以在Linux和Unix的ASM版本下工作,且不管是单实例还是RAC环境。(不能是FLEX ASM)

脚本需要以Grid软件owner的身份来运行,而且要确保pert的二进制文件来自于Oracle安装软件的home目录下。在集群环境下,这个脚本可以运行在任意节点上,在运行脚本前,请检查ASM的环境变量,确定ORACLE_SIDORACLE_HOMELD_LIBRARY_PATH设定正确,而且对于10G和11GR1版本,需要设置PERL5LIB环境变量:

export PERL5LIB=$ORACLE_HOME/perl/lib/5.8.3:$ORACLE_HOME/perl/lib/site_perl

可以以如下的方式运行脚本:

$ORACLE_HOME/perl/bin/perl find_block.pl filename block

其中:filename是要抽取的块所在的文件名,对于数据文件来说,这个文件名可以从V$DATAFILE的NAME字段获取到,block代表要从ASM抽取的块号,这个块号是数据库的块号,而不是ASM的块号。

这个脚本的输出如下:

dd if=[ASM disk path] ... of=block_N.dd

在Exadata中是这样:

kfed read dev=[ASM disk path] ... > block_N.txt

对于数据文件来说,如果文件的冗余度是external外部冗余模式,这个脚本将产生一条单一的命令,若为normal冗余,这个脚本将产生2个命令,对于high冗余,将产生3条命令。

Example with ASM version 10.2.0.1

第一个例子是单实例10.2.0.1的ASM版本,首先我在数据库中创建了一张表,插入一些数据。

[oracle@cat10g ~]$ sqlplus / as sysdba

SQL*Plus: Release 10.2.0.1.0 - Production on [date]

SQL> create table TAB1 (name varchar2(16)) tablespace USERS;

Table created.

SQL> insert into TAB1 values ('CAT');

1 row created.

SQL> insert into TAB1 values ('DOG');

1 row created.

SQL> commit;

Commit complete.

SQL> select ROWID, NAME from TAB1;

ROWID              NAME
------------------ --------------------------------
AAANE+AAEAAAAGHAAA CAT
AAANE+AAEAAAAGHAAB DOG

SQL> select DBMS_ROWID.ROWID_BLOCK_NUMBER('AAANE+AAEAAAAGHAAA') "Block" from dual;

    Block
---------
      391

SQL> select t.name "Tablespace", f.name "Datafile"
from v$tablespace t, v$datafile f
where t.ts#=f.ts# and t.name='USERS';

Tablespace   Datafile
------------ --------------------------------------
USERS        +DATA/cat/datafile/users.259.783204313

SQL>

以上我们造取了两条数据,并且定位到了数据所在的文件和BLOCK号,切换到ASM环境,注意设置正确的环境变量PERL5LIB,然后运行脚本:

$ export PERL5LIB=$ORACLE_HOME/perl/lib/5.8.3:$ORACLE_HOME/perl/lib/site_perl
$ $ORACLE_HOME/perl/bin/perl find_block.pl +DATA/cat/datafile/users.259.783204313 391
dd if=/dev/oracleasm/disks/ASMDISK01 bs=8192 count=1 skip=100359 of=block_391.dd
$

find_block.pl脚本如预期产生了输出,由于这是一个外部冗余的磁盘组,这个脚本只产生了一行dd命令的输出,我们把输出的dd命令复制后执行:

$ dd if=/dev/oracleasm/disks/ASMDISK01 bs=8192 count=1 skip=100359 of=block_391.dd
$

执行后会将块的内容输出到文本文件block_3237.dd中,然后使用操作系统的OD工具,可以看到插入表中的数据:

$ od -c block_391.dd | tail -3
0017740 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 , 001
0017760 001 003 D O G , 001 001 003 C A T 001 006 u   G
0020000
$

非常好,正是我们插入的数据!

Example with ASM version 12.1.0.1 in Exadata

在Exadata中我们不能使用dd命令抽取数据块,因为ASM的磁盘对于数据库的server来说是不可见的,为了获得数据块,我们可以使用kfed工具,因此find_block.pl脚本做了这种自适应,如果是Exadata的环境,会使用kfed工具从ASM磁盘中抽取块。

我们来看一个ASM 12.1.0.1版本下的例子,是一个Exadata环境下双节点的RAC,数据文件是PDB中的一个数据文件。

和上面的例子一样,首先创建一张表然后插入一些数据:

$ sqlplus / as sysdba

SQL*Plus: Release 12.1.0.1.0 Production on [date]

SQL> alter pluggable database BR_PDB open;

Pluggable database altered.

SQL> show pdbs

CON_ID CON_NAME OPEN MODE   RESTRICTED
------ -------- ----------- ----------
       2 PDB$SEED READ ONLY   NO
...
       5 BR_PDB   READ WRITE  NO

SQL>

$ sqlplus bane/welcome1@BR_PDB

SQL*Plus: Release 12.1.0.1.0 Production on [date]

SQL> create table TAB1 (n number, name varchar2(16)) tablespace USERS;

Table created.

SQL> insert into TAB1 values (1, 'CAT');

1 row created.

SQL> insert into TAB1 values (2, 'DOG');

1 row created.

SQL> commit;

Commit complete.

SQL> select t.name "Tablespace", f.name "Datafile"
from v$tablespace t, v$datafile f
where t.ts#=f.ts# and t.name='USERS';

Tablespace Datafile
---------- ---------------------------------------------
USERS      +DATA/CDB/054.../DATAFILE/users.588.860861901

SQL> select ROWID, NAME from TAB1;

ROWID              NAME
------------------ ----
AAAWYEABfAAAACDAAA CAT
AAAWYEABfAAAACDAAB DOG

SQL> select DBMS_ROWID.ROWID_BLOCK_NUMBER('AAAWYEABfAAAACDAAA') "Block number" from dual;

Block number
------------
       131

SQL>

同样获得插入数据的文件号和块号,切换到ASM的环境,然后运行perl脚本:

$ $ORACLE_HOME/perl/bin/perl find_block.pl +DATA/CDB/0548068A10AB14DEE053E273BB0A46D1/DATAFILE/users.588.860861901 131
kfed read dev=o/192.168.1.9/DATA_CD_03_exacelmel05 ausz=4194304 aunum=16212 blksz=8192 blknum=131 | grep -iv ^kf > block_131.txt
kfed read dev=o/192.168.1.11/DATA_CD_09_exacelmel07 ausz=4194304 aunum=16267 blksz=8192 blknum=131 | grep -iv ^kf > block_131.txt

我们观察到,find_block.pl脚本这次产生了2个命令,因此我们可以知道这是一个normal冗余的磁盘组,我们运行其中一个命令:

$ kfed read dev=o/192.168.1.9/DATA_CD_03_exacelmel05 ausz=4194304 aunum=16212 blksz=8192 blknum=131 | grep -iv ^kf > block_131.txt
$

我们将块的内容输出到了文本文件block_131.txt中,然后看到了我上面插入的数据DOG和CAT:

$ more block_131.txt
...
FD5106080 00000000 00000000 ...  [................]
      Repeat 501 times
FD5107FE0 00000000 00000000 ...  [........,......D]
FD5107FF0 012C474F 02C10202 ...  [OG,......CAT..,-]
$

Find any block

find_block.pl用来从ASM磁盘组中的任何一个文件中抽取块,不仅仅是数据文件,我对控制文件和控制文件上一个随机的块运行这个脚本:

$ $ORACLE_HOME/perl/bin/perl find_block.pl +DATA/CDB/CONTROLFILE/current.289.843047837 5
kfed read dev=o/192.168.1.9/DATA_CD_10_exacelmel05 ausz=4194304 aunum=73 blksz=16384 blknum=5 | grep -iv ^kf > block_5.txt
kfed read dev=o/192.168.1.11/DATA_CD_01_exacelmel07 ausz=4194304 aunum=66 blksz=16384 blknum=5 | grep -iv ^kf > block_5.txt
kfed read dev=o/192.168.1.10/DATA_CD_04_exacelmel06 ausz=4194304 aunum=78 blksz=16384 blknum=5 | grep -iv ^kf > block_5.txt
$

我们注意到脚本正确计算出控制文件的block size(不同于数据块的大小8K,为16K),并且脚本产生了3个不同的命令,虽然磁盘组DATA是normal冗余,但是控制文件却做了high冗余,也就是做了三副本,控制文件在这一点上跟ASM的元数据文件一样。

Conclusion

find_block.pl脚本通过dd或者kfed命令从ASM磁盘组的文件中抽取块,可能大多数情况下,我们想要从数据文件中抽取一个块,但是这个脚本不仅仅适用于数据文件,也可以从控制文件、日志文件、任何的ASM文件中抽取块。

如果文件是external外部冗余的,那么这个脚本将输出一个单一的命令,执行这个命令可以直接从ASM的磁盘中抽取块。

如果文件是normal冗余,这个脚本将输出2个命令,它用来从不同的磁盘中抽取块,这可能会比较有用,例如后台日志提示数据块损坏,ASM不能修复它,那么就可以通过镜像块来修复。

如果文件是high冗余,这个脚本将产生3个命令。

最后,使用这个脚本你不用知道文件的冗余度、块的大小,和任何其它属性,你只需要关心文件名和块号。

附脚本

#!$ORACLE_HOME/perl/bin/perl -w
#
# The find_block.pl constructs the command(s) to extract a block from ASM.
# For a complete info about this script see ASM Support Guy blog post:
# http://asmsupportguy.blogspot.com/2014/10/find-block-in-asm.html
#
# Copyright (C) 2014 Bane Radulovic
#
# This program is free software: you can redistribute it and/or modify it under
# the terms of the GNU General Public License as published by the Free Software
# Foundation, either version 3 of the License, or any later version.
# This program is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details
# at http://www.gnu.org/licenses/.
#
# Version 1.00, Oct 2014
# The initial release.
#
# Version 1.01, Oct 2014
# Minor improvements.
#
# Version 1.02, Oct 2014
# Added support for AFD disks.
#
# Version 1.03, Nov 2014
# Added sanity checks, e.g. if the requested block is reasonable,
# if the specified filename is valid, etc.
#
# Version 1.04, Nov 2014
# Improved the check for Exadata storage cell based disk.
#
use strict;
use DBI;
use DBD::Oracle qw(:ora_session_modes);
use POSIX;
# Handle the version query
die "find_block.pl version 1.04\n"
 if ( $ARGV[0] =~ /^-v/i );
# Check the number of input arguments
die "Usage: \$ORACLE_HOME/perl/bin/perl find_block.pl filename block\n"
 unless ( @ARGV == 2 );
# Get the filename from the first input argument
my $filename = shift @ARGV;
# Check if the filename makes sense.
# The 'minimum' filename is +DGNAME/filename,
# i.e. it has to begin with the '+' followed by a disk group name,
# followed by at least one '/', followed by directory or file name...
die "Error: The $filename is not a valid file name.\n"
 unless ( $filename =~ /^\+\w/ && $filename =~ /\/\w/ );
# Get the disk group name out of the user specified filename
my $diskgroup_name = substr($filename, 1, index($filename, "/") -1 );
# Get the ASM file name out of the user specified filename
my $asmfile = substr($filename, rindex($filename, "/") +1 );
# Get the block number from the second input argument
my $block_number = shift @ARGV;
# Check if the block number is an integer
die "Usage: \$ORACLE_HOME/perl/bin/perl find_block.pl filename block\n"
 unless ( $block_number =~ /^\d+$/ );
# Check if the ASM SID is set
die "Error: ASM SID not set.\n"
 unless ( $ENV{ORACLE_SID} =~ /\+ASM/ );
# Connect to the (local) ASM instance
my $dbh = DBI->connect('dbi:Oracle:', "", "", { ora_session_mode => ORA_SYSDBA })
 or die "$DBI::errstr\n";
# Check if the disk group exists and if it is mounted
my $group_number = &asm_diskgroup("group_number", $diskgroup_name);
die "Error: Disk group $diskgroup_name not mounted or does not exist.\n"
 unless ( $group_number );
# Check if the user specified file exists in the disk group
my $file_number = &asm_alias("file_number", $asmfile, $group_number);
die "Error: File $asmfile does not exist in disk group $diskgroup_name.\n"
 unless ( $file_number );
# Get the block size for the file
my $block_size = &asm_file("block_size", $group_number, $file_number);
# Get the number of blocks in the file
my $file_blocks = &asm_file("blocks", $group_number, $file_number);
# Check if the user specified block number makes sense
die "Error: Block range for file $asmfile is: 0 - $file_blocks.\n"
 unless ( $block_number >= 0 && $block_number <= $file_blocks );
# Get the disk group AU size
my $au_size = &asm_diskgroup("allocation_unit_size", $diskgroup_name);
# Work out the blocks per AU and the virtual extent number
my $blocks_per_au = $au_size/$block_size;
my $xnum_kffxp = floor($block_number/$blocks_per_au);
# Get the disk and AU numbers into the @disk_au array
my @disk_au = &asm_kffxp($file_number, $group_number, $xnum_kffxp);
die "Could not get any disk and AU numbers for file $asmfile.\n"
 unless ( @disk_au );
# Get the disk path(s) and generate the block extract command(s)
while ( @disk_au ) {
 # Do not assume anything
 my $storage_cell = "FALSE";
 # Get the disk number from @disk_au
 my $disk_number = shift @disk_au;
 # Get the AU number from @disk_au
 my $au_number = shift @disk_au;
 # Get the path for that disk number
 my $path = &asm_disk("path", $group_number, $disk_number);
 # If there is no path move to the next disk
 if ( ! $path ) {
  next;
  }
 # If ASMLIB is in use, the path will return ORCL:DISKNAME.
 # Set the path to /dev/oracleasm/disks/DISKNAME
 elsif ( $path =~ /ORCL:(.*)/ ) {
  $path = "/dev/oracleasm/disks/".$1;
  }
 # If ASM Filter Driver (AFD) is in use, the path will return AFD:DISKNAME.
 # Get the actual path from /dev/oracleafd/disks/DISKNAME
 elsif ( $path =~ /AFD:(.*)/ ) {
  if ( ! open AFDDISK, "/dev/oracleafd/disks/".$1 ) { next }
  else { chomp($path = <AFDDISK>) }
  }
 # For Exadata storage cell based disk, the path will start with o/IP address
 elsif ( $path =~ /^o\/\d{1,3}\./ ) {
  $storage_cell = "TRUE";
  }
 if ( $storage_cell eq "TRUE" ) {
  # Construct the kfed command for Exadata storage cell based disk
  # dev=$path ausz=$au_size aunum=$au_number blksz=$block_size blknum=$block_number
  # The grep filters out the kfed stuff
  print "kfed read dev=$path ausz=$au_size aunum=$au_number blksz=$block_size blknum=$block_number | grep -iv ^kf > block_$block_number.txt\n";
  }
 else {
  # Construct the dd command
  # if=$path bs=$block_size count=1 skip=$skip of=block_$block_number.dd
  my $skip=$au_number*$blocks_per_au + $block_number%$blocks_per_au;
  print "dd if=$path bs=$block_size count=1 skip=$skip of=block_$block_number.dd\n";
  }
 }
# We are done. Disconnect from the (local) ASM instance
$dbh->disconnect;
# Subs
# Get a column from v$asm_file for a given group number and file number
sub asm_file {
 my $col = shift @_;
 my $group_number = shift @_;
 my $file_number = shift @_;
 my $sql = $dbh->prepare("select $col from v\$asm_file where group_number=$group_number and file_number=$file_number");
 $sql->execute;
 my $col_value = $sql->fetchrow_array;
 $sql->finish;
 return $col_value;
 }
# Get a column from v$asm_alias for a given (file) name and group number
sub asm_alias {
 my $col = shift @_;
 my $name = shift @_;
 my $group_number = shift @_;
 my $sql = $dbh->prepare("select $col from v\$asm_alias where lower(name)=lower('$name') and group_number=$group_number");
 $sql->execute;
 my $col_value = $sql->fetchrow_array;
 $sql->finish;
 return $col_value;
 }
# Get a column from v$asm_diskgroup for a given disk group name
sub asm_diskgroup {
 my $col = shift @_;
 my $name = shift @_;
 my $sql = $dbh->prepare("select $col from v\$asm_diskgroup where name=upper('$name')");
 $sql->execute;
 my $col_value = $sql->fetchrow_array;
 $sql->finish;
 return $col_value;
 }
# Get a column from v$asm_disk for a given group number and disk number
sub asm_disk {
 my $col = shift @_;
 my $group_number = shift @_;
 my $disk_number = shift @_;
 my $sql = $dbh->prepare("select $col from v\$asm_disk where group_number=$group_number and disk_number=$disk_number");
 $sql->execute;
 my $col_value = $sql->fetchrow_array;
 $sql->finish;
 return $col_value;
 }
# Get the disk and AU numbers from x$kffxp for a given virtual extent number.
# This will return one row for an external redundancy file,
# two rows for a normal redundancy and three rows for a high redundancy.
# Well, it will return an array with disk and AU pairs, not rows.
sub asm_kffxp {
 my $file_number = shift @_;
 my $group_number = shift @_;
 my $xnum = shift @_;
 # The @disk_au array to hold the disk number, AU number rows
 my @disk_au;
 my $sql = $dbh->prepare("select disk_kffxp, au_kffxp from x\$kffxp where number_kffxp=$file_number and group_kffxp=$group_number and xnum_kffxp=$xnum");
 $sql->execute;
 # Expecting one disk number and one AU number per row
 while ( my @row = $sql->fetchrow_array) {
  # Add each (element of the) row to @disk_au array
  foreach ( @row ) { push @disk_au, $_ }
  }
 $sql->finish;
 return @disk_au;
 }
阅读更多
文章标签: 数据库 oracle asm
个人分类: oracle
想对作者说点什么? 我来说一句

没有更多推荐了,返回首页

关闭
关闭
关闭