SQL Server Performance

SQL-2000EE cluster on WIN2k3 database error

Discussion in 'SQL Server Clustering' started by desichap, May 9, 2005.

  1. desichap New Member

    Hi,

    I recently installed SQL 2000 EE on win2k3 cluster on two as a failover cluster.
    SQL fails over to the other node if I shut down any of the nodes or physically take the node offline.

    But if I try to move the group or the try to initiate a failure in cluster administrator the failover does not happen and it comes back to the current node it self after trying to go the other node.

    My concern is during either of the case that is whether fail over happens or not the database is not accessible During the time of transition.

    Could somebody guide me how to go about this as this does not meet the purpose of failover as there is no redundancy in database connectivity. I have already tried reinstalling the whole setup but no improvements.

    I use a shared disk accessible by both nodes,win2k3 with sp1 and SQL 2000 with sp3.

    Thanks
  2. Argyle New Member

    What does the OS event log says. You can usually find the cause of the non-working failover there.
  3. desichap New Member

    I have checked the event logs but there were no errors reported except for the database connection error that is during the failover. It just gives me a number and no much detail. I tried to check details with respect to that number but nothing helpful turned out with reference to the problem I am having.

    Is it normal to have a database connection error (if so I do not understand where is the redundancy and high availability) at the time fail over I am asking this as this the first time I have installed a cluster.

    I will go thru the event log again and see if I can spot any thing (I doubt)

    Thanks
  4. Argyle New Member

    Have a look at the cluster log if there is no info in the OS event logs:
    http://www.microsoft.com/resources/...server/reskit/en-us/distrib/dsdg_icl_NNTI.asp

    Any client connected to the cluster will loose connection. The failover procedure is the same as stopping and starting the sql service on a single server. Only difference is that it starts up on the other node.

    With web traffic this is usually no problem since it's in a disconnected state between requests anyway. With VB/Java or other applications with a constant connection to SQL server you need to handle this "restart" in the client (the same way you would have to do on a single server). Usually you just reconnect the application.

    INF: SQL Virtual Server Client Connections Must be Controlled by Clients
    http://support.microsoft.com/default.aspx?scid=kb;en-us;273673

    The redundancy and high availibility here is that you can take down one node while keeping the services running on the other one. This could be either for planned maintenance, reboots, hardware changes, security patches etc or an unplanned crash of a node.
  5. desichap New Member

    Hi,

    Thanks for your advice. I deleted the msdtc from the group and the failover started working fine. I was using a vb aplication to test which I modified for reconnetion and the it too is working fine reconneting within less then a minute.

    Thanks once again for your time and help.

Share This Page