SQL Server Performance

Weird LiteSpeed behaviour

Discussion in 'Third Party Tools' started by thomas, May 9, 2006.

  1. thomas New Member

    On various servers (Windows 2000 Std SP4 + SQL Server 2000 Std SP3a + hotfixes) I am seeing very weird behaviour with SQLLiteSpeed 2005 (v.4.0.2.9).

    When the first transaction log backup of the morning runs, at about 3am, LiteSpeed is backing up the entire database to the log file backup. This is using the xp_backup_log command with parameters (backup file name is constructed with the date & time in it).

    This backup follows normal overnight maintenance (full db backup, dbcc checkdb, selective reindexing). I reckon I'm going to have to seek support from Quest, but I wondered if any of you have seen this behaviour?

    There is no way anything other than an xp_backup_log is being executed, but it is definitely resulting in a full database backup as opposed to a t-log backup. And this is happening on a range of different servers.
  2. satya Moderator

    Tom
    Have you got in touch with Vendor in this case?

    Satya SKJ
    Microsoft SQL Server MVP
    Contributing Editor & Forums Moderator
    http://www.SQL-Server-Performance.Com
    This posting is provided “AS IS” with no rights for the sake of knowledge sharing.
  3. thomas New Member

    No I haven't yet, but I'm about to. It occurs on databases which, during maintenance, are set to BULK_LOGGED and have reindexing run on them, then get set back to FULL. This shouldn't require a full backup after to keep the transaction logs consistent, but it seems like SQLLiteSpeed somehow determines that that's necessary.

    But the "full backup" produced, although it's the same size as a full database backup, can't be restored from - so it's pretty useless. I'll let you know what Quest say.
  4. satya Moderator

    Then there must be some sort of non-logged operation occuring between the recovery model switching.

    Satya SKJ
    Microsoft SQL Server MVP
    Contributing Editor & Forums Moderator
    http://www.SQL-Server-Performance.Com
    This posting is provided “AS IS” with no rights for the sake of knowledge sharing.
  5. thomas New Member

    Well reindexing is the only thing. I have recreated this issue on a small database. But reindexing during a bulk-logged state shouldn't mean a full backup is required before continuing log backups. In previous situations I've done this with LiteSpeed (earlier version though, 2.6 something) without a problem.

    Waiting for Quest to call me back...
  6. thomas New Member

    The newest version of LiteSpeed (4.5.0.00199) fixed this issue. Anyone using earlier versions of LiteSpeed 2005 beware, your t-logs may not be recoverable if they are preceeded by bulk-logged operations as was the case in our scenario.

    I've just spent the whole day installing the new version on all our servers! *sigh*
  7. satya Moderator

    Tom
    Thanks for the Update, I'm sure that will help others as Quest products are popular.

    Satya SKJ
    Microsoft SQL Server MVP
    Contributing Editor & Forums Moderator
    http://www.SQL-Server-Performance.Com
    This posting is provided “AS IS” with no rights for the sake of knowledge sharing.

Share This Page