VCS Retrying resource before failover - Veritas Cluster Server

This is a discussion on VCS Retrying resource before failover - Veritas Cluster Server ; I am interested in knowing whether VCS is capable of trying to re-start a resource before failing to another node. Very often we will have an application crash, like oracle or pcnetlink and the node and OS are perfectly fine. ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: VCS Retrying resource before failover

  1. VCS Retrying resource before failover


    I am interested in knowing whether VCS is capable of trying to
    re-start a resource before failing to another node. Very often
    we will have an application crash, like oracle or pcnetlink
    and the node and OS are perfectly fine. In this case I would
    like VCS to try an re-start the application rather than move the
    entire resource group to another node. Has anyone ever tried
    this and is i possible. I visited with Tech support and they
    did not seem to think it was.

  2. Re: VCS Retrying resource before failover


    "Greg Potter" wrote:
    >
    >I am interested in knowing whether VCS is capable of trying to
    >re-start a resource before failing to another node. Very often
    >we will have an application crash, like oracle or pcnetlink
    >and the node and OS are perfectly fine. In this case I would
    >like VCS to try an re-start the application rather than move the
    >entire resource group to another node. Has anyone ever tried
    >this and is i possible. I visited with Tech support and they
    >did not seem to think it was.


    YES. If you configure the OnlineRetryInterval to a non zero number. Eg. 2
    will make VCS try 2 times before failing over. Hope this helps. For more
    info. look at the VCS_users guide.

  3. Re: VCS Retrying resource before failover

    Greg Potter wrote:
    >
    > I am interested in knowing whether VCS is capable of trying to
    > re-start a resource before failing to another node. Very often
    > we will have an application crash, like oracle or pcnetlink
    > and the node and OS are perfectly fine. In this case I would
    > like VCS to try an re-start the application rather than move the
    > entire resource group to another node. Has anyone ever tried
    > this and is i possible. I visited with Tech support and they
    > did not seem to think it was.


    The VCS framework has the ability to restart resources before returning
    offline to the engine. to increase the restart limit to 3 use the
    following command:

    #haconf -makerw
    #hatype -modify RestartLimit 3
    #haconf -dump -makero

    These changes will go in to the types.cf file and will effect all agents
    of that type, so if you set restart to 3 for Oracle all Oracle type
    resources will restart 3 times.


    From the Agent developer's guide 1.3.X:

    RestartLimit
    Affects how the agent responds to a resource fault (see
    FaultOnMonitorTimeouts on
    page 63 and ToleranceLimit on page 66). A non-zero RestartLimit causes
    VCS to invoke
    the online entry point instead of failing over the service group to
    another system. VCS
    attempts to restart the resource according to the number set in
    RestartLimit before it gives
    up and fails over. However, if the resource remains online for the
    interval designated in
    ConfInterval, earlier attempts to restart are not counted against
    RestartLimit. Default is 0.

    Note The agent will not restart a faulted resource if the clean entry
    point is not
    implemented. Therefore, the value of the RestartLimit parameter applies
    only if
    clean is implemented.
    --
    Bill Borsari
    VERITAS Software
    support.veritas.com

+ Reply to Thread