5 Replies Latest reply on Dec 14, 2012 3:50 PM by Henrie Cuijpers

    Problems maybe with async

    Henrie Cuijpers
      Hi all,

      i have a strange problem. Yesterday i activated async io (linux) by setting FILESYSTEMIO_OPTIONS to setall and bouncing the instance. Everything looked normal until this morning the backup came along...

      The backup exited with a ORA-19599: block number 2048 is corrupt in archived log "filename".

      When i saw this i imeddiatly took a backup with RMAN> backup database; wich succesfully completed
      After that i deleted (move out of the way) all archived logs and issued a crosscheck and delete obsolete.
      Then again i did a full backup, but now inlcuding the archived logs.
      Again there was a ora-19599 for block 2048 but now in a recent log....

      I repeated these steps 3(!) times and i was able to reproduce the same behavior exactly...

      After that i disabled the async io (because that was the only thing i changed) and repeated the steps once agian..... This time: no errors!!!

      So it looks like the async io corrupts my archived redo logs??? Anyone some experience on this?

      btw it is a on 64 bit CentOS Linux release 6.0 (Final) ( kernel 2.6.32-71.el6.x86_64 )

      my hpacucli tells me all disks are OK

      Edited by: hcHoep on Dec 14, 2012 2:07 PM