Query running in between the failover

  • if one query is running and in between cluster failover happens then what will happen to the query?

  • It will result in a connection failure and the query will stop running as part of the controlled shutdown of services.

    All services are shutdown on one node and restarted on the new node.

  • Msg 121, Level 20, State 0, Line 0

    A transport-level error has occurred when receiving results from the server. (provider: TCP Provider, error: 0 - The semaphore timeout period has expired

    In a cluster environment,when 1 node down then services start on another node

    Regards,

    Syed Jahanzaib Bin Hassan

    MCTS | MCITP | OCA | OCP | OCE | SCJP | IBMCDBA

    My Blog

    http://www.aureus-salah.com

    Regards,
    Syed Jahanzaib Bin Hassan
    BSCS | MCTS | MCITP | OCA | OCP | OCE | SCJP | IBMCDBA

    My Blog
    www.aureus-salah.com

  • Thats a good one.

    Yes it is very true cluster failover is the SQL service fail over, hence you will receive this error.

    One thing I want to know : Is MS doing some thing for session continuation as that is taken care in Oracle RAC.

    How does the DB buffer react during the fail over? So all the commited info goes is there any time for checkpoint to fire?

    "More Green More Oxygen !! Plant a tree today"

  • Minaz Amin (5/16/2011)


    One thing I want to know : Is MS doing some thing for session continuation as that is taken care in Oracle RAC.

    No idea, what does RAC do?

    When the one node fails, any connections to it are dropped. They have to reconnect and they'll get the other node.

    How does the DB buffer react during the fail over? So all the commited info goes is there any time for checkpoint to fire?

    Depends on the type of failure. If it's a controlled shutdown of the one node then there will be time for a checkpoint. If there's a hardware failure, not so much. When SQL restarts on the other node it will run crash recovery on all or its databases.

    Gail Shaw
    Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
    SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability

    We walk in the dark places no others will enter
    We stand on the bridge and no one may pass
  • RAC is session persistence then user will not interrupt when node move,RAC is also for ACTIVE/ACTIVE which means real load balancing 2 or more Servers point to the same Oracle instance.if RAC have 2 nodes then load look like 50 50 % on each node if one fail then 100 % on second node

    NLB Server (Network Load Balancing) by Microsoft

    http://technet.microsoft.com/en-us/library/cc738078(WS.10).aspx

    check this article for session persistence at the side of application IIS

    Regards,
    Syed Jahanzaib Bin Hassan
    BSCS | MCTS | MCITP | OCA | OCP | OCE | SCJP | IBMCDBA

    My Blog
    www.aureus-salah.com

  • Practically when a cluster resource fails to another node applications will reconnect depending upon the design, I supported one app where a failover resulsted in all users getting locked out. There was no facility to block unlock the accounts which wasn't very clever when you have 650 user lcoked out! In this sort of situation you would say that the applciation is not cluster aware. As far as transactions on the server are concerned where ever possible these will rollback, in reality if your application is cluster aware most users are unaware of a failover, they may get an error but they will reconnect automatically and carry on. ( it's possible that some work may be lost of course )

    To be honest it does depend upon the application and n-tier type apps, in my experience, tend to handle this better.

    I do think if you're going to work with clustered SQL Server you should understand windows clustering, the cluster team have an excellent blog and there are some excellent books on the subject.

    As far back as I can remember there have been a constant stream of posts talking about load balancing of sql server, which isn't what windows clustering does, there's more than adequate coverage in technet and msdn which is freely available, reading through these first would avoid the repetition of these questions and the wealth of disinformation too.

    Microsoft also run some excellent training courses.

    [font="Comic Sans MS"]The GrumpyOldDBA[/font]
    www.grumpyolddba.co.uk
    http://sqlblogcasts.com/blogs/grumpyolddba/

Viewing 7 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic. Login to reply