Iniciar sesión
 
InicioBibliotecaAprendizajeDescargaSoporteComunidadForos
Recursos para Profesionales de TI >
> 
> 
Windows 2008 Cluster Failover behavior

Answered Windows 2008 Cluster Failover behavior

  • lunes, 07 de mayo de 2012 7:50
     
     
    Inicie sesión para votar
    0

    We conducted a Failover drill to test the failover behavior of a 2-node Windows 2008 Cluster.  In the setup, there are 2 Windows 2008 Enterprise servers with SP2.

    In the cluster, there is a shared drive service and a clustered IBM DB2 application service.

    The test :

    All the network cables, including the Heartbeat cable were unplugged at the active node. 

    Expected Result :

    We expected that there will be short interruption while the DB2 service and the file system to fileover to the standby server. 

    Actual Result :

    Service is interrupted and could not be recovered by failover from active node to standby node.

    Question

    Should Windows 2008 cluster manager be able to handle the above scenario that all network connections including the heart beat cable are removed at the active node? 

    Thanks for your advice.

    Adolphus

     

Todas las respuestas

  • lunes, 07 de mayo de 2012 9:09
     
     Respuesta propuesta
    Inicie sesión para votar
    0

    Is there any errors showing in cluster validation report?

    Please try a normal shutdown of active node to see if the fail overs happen as expected.

    Check out these links:

    http://support.microsoft.com/kb/197047 (Increase the group failover count, if required)

    http://blogs.technet.com/b/vipulshah/archive/2009/06/17/failover-cluster-testing-methods.aspx

    http://technet.microsoft.com/en-us/library/cc757030%28v=ws.10%29.aspx

    - thestriver


    • Editado thestriver lunes, 07 de mayo de 2012 9:21
    • Propuesto como respuesta thestriver viernes, 11 de mayo de 2012 1:19
    •  
     
  • lunes, 07 de mayo de 2012 10:50
     
     
    Inicie sesión para votar
    0
    Hello Instaed of removing LAN cable, stop the cluster service in one node and test for the fail over whether it happens or not?

    Regards, Ravikumar P

     
  • lunes, 07 de mayo de 2012 11:44
     
     
    Inicie sesión para votar
    0
    Also check event viewer of standby server and check cluster logs for more details 

    Regards, Ravikumar P


     
  • miércoles, 09 de mayo de 2012 10:10
     
     Respondida
    Inicie sesión para votar
    0

    Can you still contact/manage the cluster? I wasn't aware that DB2 was a cluster-aware application.... You should still be able to manage the cluster in this scenario however DB2 may be offline (if it is not a truely cluster-aware application). If you cannot connect to the cluster (using Failover Cluster Manager) then that is a different issue.

    • Propuesto como respuesta thestriver viernes, 11 de mayo de 2012 1:19
    • Marcado como respuesta Vincent HuModerator viernes, 11 de mayo de 2012 5:04
    •