Re: [suse-oracle] Detecting oracle deadlocks in 8i and 9i

From: Martin Berg (martin@berg.dk)
Date: Fri Oct 26 2001 - 12:12:33 PDT

  • Next message: Marquez, Chris: "[suse-oracle] SQLPLUS "SYS/manager9i@orcl AS SYSDBA"...shutdown, but NOT startu p????"

    Message-ID: <3BD9B5A1.8010504@berg.dk>
    Date: Fri, 26 Oct 2001 21:12:33 +0200
    From: Martin Berg <martin@berg.dk>
    Subject: Re: [suse-oracle] Detecting oracle deadlocks in 8i and 9i
    

    Matt wrote:

    >Hi Martin and All,
    >
    >What do all the select statements have to have a select for update
    >*nowait* on them if I am inserting data or deleting from one table to another ?
    >
    >Or should it be only when creating views ?
    >Or when altering (dropping/creating) partitions ?
    >
    >I think most of the problem is a select is happening for a creation of a
    >view whilst a sqlloader session is in operation.
    >
    >Are there any fantastic documentations on the area of deadlocking ?
    >
    Not really from Oracle - Steves and my previous answers covers most what can
    be said about deadlocks.

    >Also is deadlocking handeled better in 8.1.7 and 9i then on 8.1.6 ?
    >
    Deadlocks are not dependent of the version - it is purely a question of
    what
    sequence objects (normally rows or tables) are being locked by multiple
    database sessions - and (important): it is the application logic and
    the users
    usage of the applications that gives the deadlock.

    I must admit though that by your question I'm starting to wonder if what
    you refer to as deadlocks are the same as deadlocks as defined by Oracle.

    Could you describe a little bit about the situations where you encounter
    deadlocks?

    Martin Berg

    ---------------------------------------------------------------------
    To unsubscribe, e-mail: suse-oracle-unsubscribe@suse.com
    For additional commands, e-mail: suse-oracle-help@suse.com



    This archive was generated by hypermail 2.1.0 : Fri Oct 26 2001 - 12:10:36 PDT