ASM support utilities

Some examples and comments on the use of ASM support utilies, in particular kfed and amdu. These utilities allow troubleshooting of ASM diskgroups also with unmounted diskgroups. Beware, kfed in write mode can corrupt your ASM config too. Note: when ASM diskgroups can be mounted similar info are available in X$ tables as shown in ASM_Internals. Oracle support note ID 553639.1 has useful info on amdu and kfed.


  • Allows to dump ASM contents without opening diskgroups, allows to check ASM file mirroring when using normal redundancy
  • powerful tool for troubleshooting, introduced in 11g, usable in 10g too.
  • amdu run creates a directory with a report.tx file, plus optionally the and the .map and .img dump files (for metadata and data, the actual output depends on the switches specified in the command line )
  • a few examples of interest:

# displays online help for the utility
  $ amdu -help   

# extracts file 267 from ASM diskgroup TEST4_DATADG1 
# Note: works as asmcmd cp but olso on dismounted disk groups!
  $ amdu -dis '/dev/mapper/itsto*p1' -extract TEST4_DATADG1.267  

# compares primary and mirror extents in normal redundancy disk groups 
# Useful to check for potential corruption issues 
# results in the report.txt file
  $ amdu -dis '/dev/mapper/itsto*p1' -compare -extract TEST4_DATADG1.267 -noextract 

# dump contents of a given diskgroup and does not create image file
# the .map file reports on all files found (column number 5 prefixed by the letter F)
  $ amdu -dis '/dev/mapper/itsto*p1' -noimage -dump TEST4_DATADG1

# print 10 blocks for the first extent of file 267
  $ amdu -dis '/dev/mapper/itsto*p1' -print TEST4_DATADG1.F267.X1.B1.C10

Example: how to retrieve DB files from a dismounted diskgroup
  • extract file 256 from the data diskgroup as it often the first copy of controlfile
  • (or use a diskgroup dump as above to find files list. db alert log and RMAN catalog are other sources to check)
  • use strings command to find the list of files to be retrived and extract them with amdu
  • -fullscan, -baddisks and/or -former can be of help for some cases
     $ amdu -dis '/dev/mapper/itsto*p1' -extract TEST4_DATADG1.256
     $ strings TEST4_DATADG1.256|grep TATEST4_DATADG1|less # find list of <FILENUM>
     $ amdu -dis '/dev/mapper/itsto*p1' -extract TEST4_DATADG1.<FILENUM> 


  • kfed can be used to read and write ASM metadata. In particular disk headers and ASM (hidden) metadata files.
  • note: kfed in write mode is a powerful but potentially dangerous tool
# displays online help for the utility
  $ kfed -help  

# reads the disk header to stdout
  $ kfed op=read dev=/dev/mapper/itstor741_11p1  

# reads the specified AU and block into file /tmp/a
  $ kfed op=read dev=/dev/mapper/itstor741_11p1 aunum=3 blknum=3 text=/tmp/a 

# writes from /tmp/a into the specified AU and block 
#block checksum is computed and written together with data
  $ kfed op=write dev=/dev/mapper/itstor741_11p1 aunum=3 blknum=3 text=/tmp/a 

Example: how to change an ASM diskgroup parameter on a dismounted disk group
  • the parameter is in ASM (hidden) file #9
  • find AUs that contain that information (can be up to 3 mirrored copies in a normal redundancy diskgroup with 3 or more failgroups)
  • read current value into a flat file, update the file and write the updated file with checksum using kfed
  • dismount and mount diskgroup to see change (query v$asm_attribute)

-------------- ----------------------------------------
            41 /dev/mapper/itstor741_11p1
            22 /dev/mapper/itstor739_10p1
            34 /dev/mapper/itstor740_9p1

READ: kfed read /dev/mapper/itstor741_11p1 aunum=3 blknum=3 text=/tmp/a
EDIT: vi /tmp/a  and change attribute value (for example change smart_scan_enable from TRUE to FALSE 
         to disable exadata smart scans in this particular diskgroup, note update length field too)
kfede[0].name:       smart_scan_capable ; 0x034: length=18
kfede[0].value:                   FALSE ; 0x074: length=5
kfede[0].length:                      5 ; 0x174: 0x0005

kfede[0].name:       smart_scan_capable ; 0x034: length=18
kfede[0].value:                    TRUE ; 0x074: length=4
kfede[0].length:                      4 ; 0x174: 0x0004

WRITE PRIMARY EXTENT: kfed write /dev/mapper/itstor741_11p1 aunum=3 blknum=3 text=/tmp/a
  kfed write /dev/mapper/itstor739_10p1 aunum=3 blknum=3 text=/tmp/a
  kfed write /dev/mapper/itstor740_9p1 aunum=2 blknum=3 text=/tmp/a

SQL> (dismount) and mount the group again to see the chagen in v$asm_attribute
Note after changing the parameter you may need to explicitly turn smart scans off on the RDBMS
alter system set cell_offload_processing=FALSE scope=both sid='*';

Example: how to undrop ASM diskgroup and/or recover files from dropped diskgroups
  • method 1: use amdu to recover the files: amdu -dis '/dev/mapper/devstor4_1p1' -former -extract RDTEST2_TESTDROP.256
  • method 2: use kfed to change the disk headers back to member
    kfed read /dev/mapper/devstor4_1p1 aunum=0 blknum=0 text=devstor4_1p1.txt
    vi devstor4_1p1.txt and change
    kfdhdb.hdrsts:                        4 ; 0x027: KFDHDR_FORMER
    kfdhdb.hdrsts:                        3 ; 0x027: KFDHDR_MEMBER
    kfed write /dev/mapper/devstor4_1p1 aunum=0 blknum=0 text=devstor4_1p1

Example: how to rename an ASM disk

  • In the case of ASM 12c, the procedure is:
    • alter diskgroup .. mount restricted
    • alter diskgropu rename disk '...' to 'new path';
  • This is an undocumented and unsupported procedure that can be used in 11g (at your own risk):
    1) Identify all the mirror extents of ASM file N.2 for the relevant diskgroup (i.e. the disk directory). Example:
    select disk_kffxp,au_kffxp, path from x$kffxp x,v$asm_disk_stat v
    where x.group_kffxp=2 and number_kffxp=2
          and x.group_kffxp=v.group_number and x. disk_kffxp=v.disk_number
    order by x.pxn_kffxp;
    2) dismount the diskgroup
    3) Read the first copy of the disk header and update it with the new disk name and length. Example:
    kfed read /dev/mapper/disk1name_p1 aunum=0 blknum=0 text=disk1name_p1_au0blk0
    make a backup copy of the file disk1name_p1_au0blk0
    vi disk1name_1p1_au0blk0
    update -> kfddde[0].dskname:              NEWNAME ; 0x038: length=7
    kfed write /dev/mapper/disk1name_p1 aunum=0 blknum=0 text=disk1name_1p1_au0blk0
    Note: it does not seem to be necessary to update the other copies of the disk header, ASM seems to take care of that when we will mount the diskgroup.
    4) Use kfed to read, modify and then write the new disk name in the relevant entry of the directory. 
    This is similar to point N.3 with the notable minor differences that: 
    - From step (1) above we know in which aunum and disk we will find the directory but not necessarily in which blknum. It's just a matter of trying blknum=0, then blknum=1, etc. It won't take long.
    - For diskgroups in normal or high redundancy it's advisable to update all the mirror copies.
    5) mount the diskgroup, monitor the ASM alert log and hope for the best :)
    6) Some tips to recover from errors: 
    - dismount the diskgroup
    - copy back the previous values of the modified blocks with using from the backup files
    - mount the diskgroup
    - the affected disks will probably be offline, online them back.

How to use KFED and AMDU in 10g

kfed and amdu are available by default in 11g in $ORACLE_HOME/bin. In 10g one needs
  • For kfed a link operation needs to be run:
cd $ORACLE_HOME/rdbms/lib
make -f kfed
  • for amdu, one needs to download it from support. See note note ID 553639.1
  • in alternative an installation of and additional ORACLE_HOME with 11gR2 binaries should also work for the purposes of using amdu


BBED is an Oracle internal utility to do BLOCKEDIT. It is extensively discussed at BBED doesn't work on ASM by default. A work-around has been studied by An example to get started can be found here below.
Note that bbed does not build out of the box on 11g and 12c, as there are a few files missing in the Oracle binaries distribution to properly build bbed. There is a simple workaround, that consists on copying the relevant files from 10g binaries before running make. More details at this blog entry by Laurent Leturgez

cd $ORACLE_HOME/rdbms/lib
make -f $ORACLE_HOME/rdbms/lib/bbed

Dump an extent or an ASM file with the methods described above, for example:
amdu -dis '/dev/mapper/itsto*p1' -extract TEST4_DATADG1.555

Edit: vi bbed.par (or rather follow the example in the ASMBBED link above) 

$ORACLE_HOME/rdbms/lib/bbed parfile=bbed.par

First version, Jan 2010,
Additional examples for amdu and kfed, May 2014,

Edit | Attach | Watch | Print version | History: r15 < r14 < r13 < r12 < r11 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r15 - 2015-06-23 - LucaCanali
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    PDBService All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback