Skip to Content
avatar image
Former Member

Restore Failed due to Corrupt Backup?

Hi,

We suffered from a failure attempting to restore from a backup recently. Trying to restore (from two different backups) of our QA3 system resulted in the following error:

2012-02-15 17:16:06 4252 ERR 3 DBMKnl Received the result of an administrative statement from the database

4252 ERR 9 DBMKnl Statement: CREATE INSTANCE WITH RESTORE DATA FROM '/export/QA3/QA3_COM' FILE BLOCKSIZE 8 MEDIANAME 'FILE_DATA_BAK2'

4252 ERR 10 DBMKnl Returncode: -903

4252 ERR 5 DBMKnl Errortext: Host file I/O error

2012-02-15 17:16:06 4252 ERR -24580 DBMSrv ERR_COMMAND_FAILED: Command 'db_activate' has ended and failed with return code -24988.

4252 ERR -24988 DBMSrv ERR_SQL: SQL error

4252 ERR -24778 DBMSrv -903,Host file I/O error

4252 ERR -24778 DBMSrv 6,Data recovery failed

4252 ERR -24778 DBMSrv 11,The number of pages expected in the backup (40572007) does not match the number of pages actually recovered (40572003), 0 duplicate pages found in the backup

...the number of pages was slightly different on another attempt due to the different dates/sizes of the two backups we tried.

Has anyone seen this error before and/or know exactly what was wrong? We were restoring from an NFS mount, but a subsequent restore from a production backup using the exact same NFS mount worked fine (which points to the backup medium being corrupt rather than a problem with the NFS mount).

Let me know if any other logs are required to assist in diagnosis. I will also be logging a customer message regarding this.

Thanks,

Chaz.

Add comment
10|10000 characters needed characters exceeded

  • Follow
  • Get RSS Feed

1 Answer

  • avatar image
    Former Member
    Feb 20, 2012 at 09:33 PM

    Also note a backup verification was successful:

    2012-02-16 01:44:18 23828 INF 3 DBMKnl Received the result of an administrative statement from the database

    23828 INF 9 DBMKnl Statement: RESTORE DATA FROM '/export/QA3/QA3_COM' FILE BLOCKSIZE 8 MEDIANAME 'FILE_DATA_BAK2'

    23828 INF 10 DBMKnl Returncode: 0

    23828 INF 7 DBMKnl Data: DATE.............. 20120215

    23828 INF 7 DBMKnl Data: TIME.............. 00225043

    23828 INF 7 DBMKnl Data: SERVERDB.......... QA3

    23828 INF 7 DBMKnl Data: SERVERNODE........ msaptmp3

    23828 INF 7 DBMKnl Data: KERNEL VERSION.... Kernel 7.8.02 Build 026-121-243-456

    23828 INF 7 DBMKnl Data: PAGES TRANSFERRED. 40572088

    23828 INF 7 DBMKnl Data: PAGES LEFT........ 0

    23828 INF 7 DBMKnl Data: NO OF VOLUMES..... 1

    23828 INF 7 DBMKnl Data: MEDIA NAME........ FILE_DATA_BAK2

    23828 INF 7 DBMKnl Data: TAPE NAME......... /export/QA3/QA3_COM

    23828 INF 7 DBMKnl Data: TAPE ERRORTEXT.... (null)

    23828 INF 7 DBMKnl Data: TAPE LABEL........ DAT_000001084

    23828 INF 7 DBMKnl Data: IS CONSISTENT..... true

    23828 INF 7 DBMKnl Data: FIRST IO SEQUENCE. 28239074

    23828 INF 7 DBMKnl Data: LAST IO SEQUENCE.. (null)

    23828 INF 7 DBMKnl Data: DBSTAMP1 DATE..... 20120206

    23828 INF 7 DBMKnl Data: DBSTAMP1 TIME..... 00131432

    23828 INF 7 DBMKnl Data: DBSTAMP2 DATE..... (null)

    23828 INF 7 DBMKnl Data: DBSTAMP2 TIME..... (null)

    23828 INF 7 DBMKnl Data: BD PAGE COUNT..... 40572007

    23828 INF 7 DBMKnl Data: TAPEDEVICES USED.. 1

    23828 INF 7 DBMKnl Data: DB_IDENT.......... msaptmp3:QA3_20120206_131432

    23828 INF 7 DBMKnl Data: MAX USED DATA PNO. 0

    23828 INF 7 DBMKnl Data: CONV PAGE COUNT... 24116

    2012-02-16 01:44:18 23828 INF 419 DBMSrv Command 'recover_check' has ended with return code 0.

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member Former Member

      Hi Sahil,

      This is the log from a prior attempt that show what happened when we tried to online the DB:

      -


      2012-02-14 17:26:23 3559 ERR 3 DBMKnl Received the result of an administrative statement from the database

      3559 ERR 9 DBMKnl Statement: CREATE INSTANCE WITH RESTORE DATA FROM '/export/QA3/QA3_COM' FILE BLOCKSIZE 8 MEDIANAME 'FILE_DATA_BAK2'

      3559 ERR 10 DBMKnl Returncode: -903

      3559 ERR 5 DBMKnl Errortext: Host file I/O error

      2012-02-14 17:26:23 3559 ERR -24580 DBMSrv ERR_COMMAND_FAILED: Command 'db_activate' has ended and failed with return code -24988.

      3559 ERR -24988 DBMSrv ERR_SQL: SQL error

      3559 ERR -24778 DBMSrv -903,Host file I/O error

      3559 ERR -24778 DBMSrv 6,Data recovery failed

      3559 ERR -24778 DBMSrv 11,The number of pages expected in the backup (40587081) does not match the number of pages actually recovered (40587072), 0 duplicate pages found in the backup

      2012-02-14 17:28:55 21152 INF 1 DBMSrvCo A DBM Server client connection was established at 2012-02-14 17:28:52 (client process has process ID 21150 on computer msaptmp3.xxxx.net).

      2012-02-14 17:28:55 21152 INF 283 DBMSrv Command 'db_online' is being executed.

      2012-02-14 17:28:55 21152 INF 1 DBMKnl Sending an administrative statement to the database

      21152 INF 8 DBMKnl Statement: RESTART

      2012-02-14 17:28:57 21152 ERR 3 DBMKnl Received the result of an administrative statement from the database

      21152 ERR 9 DBMKnl Statement: RESTART

      21152 ERR 10 DBMKnl Returncode: -3070

      21152 ERR 5 DBMKnl Errortext: A previous data recovery was incomplete and must be repeated

      2012-02-14 17:29:03 3583 INF 2 DBMSrvCo A DBM Server client connection was released (client process has process ID 3559 on computer DBM Server Shared Memory).

      2012-02-14 17:29:03 21152 ERR -24580 DBMSrv ERR_COMMAND_FAILED: Command 'db_online' has ended and failed with return code -24988.

      21152 ERR -24988 DBMSrv ERR_SQL: SQL error

      21152 ERR -24778 DBMSrv -3070,A previous data recovery was incomplete and must be repeated

      21152 ERR -24778 DBMSrv 3,Database state: OFFLINE

      21152 ERR -24778 DBMSrv Internal errorcode, Error code 8052 "prev_data_recovery_was_incomplete"

      21152 ERR -24778 DBMSrv 20017,RestartFilesystem failed with 'A previous data recovery was incomplete and must be repeated'

      -


      Note we cant now recreate this scenario easily as we needed the sandbox available and therefore had to perform a refresh from production to it, which, coincidentally, worked fine using a restore from backup on the same NFS mount as the failing one.

      Thanks,

      Chaz.