SQL Server Performance

I/O Error (torn page)

Discussion in 'General DBA Questions' started by Jon M, Feb 8, 2004.

  1. Jon M Member

    Hi,

    I'm trying to recover a database that became 'SUSPECT'. I tried running the sp_resetstatus, stop and restart my SQL Server. Since that didn't work, I detached the database right away and it was not cleanly detached. I tried attaching the database back, but it's giving this below message:

    Error 823: I/O error (torn page) detected during read at offset 0x000015c4cae000 in file
    MyMDFHere.MDF

    Attaching database has failed.


    Any work-around or suggestions will be highly appreciated. My only hope is the tape backup but as much as possible I'd like to try recovering my MDF/LDF if there's any other feasible method.

    Jon M

  2. Luis Martin Moderator

  3. kay27 New Member

    Jon,
    since the database can not be accessed by SQL Server, you can not use dbcc repair commands. So I can not help you on this.
    But there may be more bad news for you since you may still have the underlying problem which caused all this. I had a similar error message (err 823 I/O error) on a single table. After intense research in forums, KB, hotlines etc., I am now convinced that the reason causing this error message is below SQL Server and below the OS layer. My best guess is write cache, disk firmware, OS disk driver or plain hardware (disk or controller) problems. Still researching.
    I had err823 on a HP ProliantDL380G3 with a pair of 72G 10K rpm disks in Raid1 with a smart array 5i+ controller, battery-backed write cach set to 50%read and50% write.
    what is your config ?

  4. ChrisFretwell New Member

    Jon, I hope you have a backup around. Drop your database, complete, full drop. Run the disk diagnositics and make sure your system is okay. Then recreate your DB from the last clean backup (I think you can use the with move option).

    Sometimes databases go suspect when there is a small hiccup and sql gives you a chance to check it out and reset the status if its okay. Sometimes suspect databases are really corrupt and unrecoverable.

    Horrible time to test your recovery procedures.
  5. Jon M Member

    I appreciate all your replies.<br /><br />kay27, I've got a pair of 200gb (raid 1) for my OS and another for my transaction log. I'm using raid 5 for my data file.<br /><br />I was trying to revive/recover the actual file by referring to BOL, resetting the suspect status. I've also considered this link below:<br /><i<a target="_blank" href=http://www.nigelrivett.net/RecoverCorruptDatabase.html>http://www.nigelrivett.net/RecoverCorruptDatabase.html</a></i><br /><br />I was close in transferring the data using this method but I finally decided to use the latest backup tape because I believe my tape will give me a more accurate result. Although, I just have to re-create other tables but not a problem there's only 3 tables. <br /><br />I'm back to business. [<img src='/community/emoticons/emotion-2.gif' alt=':D' />]<br /><br />Jon M

Share This Page