ASM disk full - metadata problem

From: De DBA <>
Date: Thu, 19 Dec 2013 10:46:49 +1000
Message-ID: <>


We are trying to get an ASM diskgroup going again after one disk ran out of space. This particular disk group contains some 20 disks, most of which are 500G, but some are 1TB, one is 50G and another - the problem disk - is 100G.

We could determine which files have chunks on the full disk and fortunately could drop some of those. Even after dropping those files though, v$asm_disk shows that no space is freed up and rebalancing still fails with ORA-15041 (space exhausted). We ran a disk group check and found this:

SQL> ALTER DISKGROUP DATA01 CHECK NOREPAIR NOTE: starting check of diskgroup DATA01 ...
Wed Dec 18 19:08:38 2013
NOTE: disk DATA01_0009, used AU total mismatch: DD={102427, 0} AT={102390, 0} ...

This is the problem disk - similar messages are issued for each disk in the diskgroup. If I interpret this correctly, the disk metadata say that the full 100G is used, but actually only 99,9GB is allocated? That is closer to what we would expect: a query on X$KFFXP shows that this file has 102343 AU mapped to files.

Obviously this has to be repaired, but just as a sanity check: is it safe to run "ALTER DISKGROUP DATA01 CHECK" to repair the disk inconsistencies? What are the risks associated with doing this, if any? Do I have to take more precautions than take a full DB backup..

Some details:

  • Oracle
  • Suse SLES.. (unknown -- kernel version
  • This is a 2-node RAC installation


Received on Thu Dec 19 2013 - 01:46:49 CET

Original text of this message