Difference between possible owners and preferred owners

  • Hi

    I have some confusion between possible owners and preferred owners in sql server clustering enviroment and other question my mind is if I want to install some service on the clusters and the best practices I would follow is avoid unplanned failover and to that what should I do? whether do I need to remove the other nodes in preferred owners or remove the nodes in possible owners list which is the one that is correct

    and the other thing is If I have a active group that a preferred owners is 1,2, 3 order but in the group the SQL server agent has possible owner order is 2,1,3 will this has any descrepancies?

    Thanks,

    Ravi

  • Possible owners are nodes which are allowed to host the service, so if there are four nodes, and nodes 1, 2, and 3 are possbile owners, the service will NEVER fail over to node 4.

    Preferred owners are just that- the node we would lke to have it on under ideal conditions, but maybe not the only one it can be on... for example, let's say node 1 and 3 are "preferred" owners, and nodes 1,2 and 3 are possible owners, then if the service is on node 1 and node 1 fails, then the service will move to node 3 and only go to node 2 if both 1 and 3 are not available.

  • to answer the second part, if properly configured, the agent is dependent upon the server service and will therefore go to the same node as the server server... so assuming the list contains the same nodes, then the order does not matter.

  • Possible owners are defined at the resource level and dictate which nodes in the Windows cluster are able to service this resource


    For instance, you have a 3 node cluster with Node A, Node B and Node C. You have a clustered disk resource "MyClusteredDisk", if you remove Node C from the possible owners of the clustered disk resource "MyClusteredDisk" then this disk will never be failed over to Node C.

    Preferred owners are defined at the resource group level and define the preferred node ownership within the Windows cluster


    For instance, you have a 3 node cluster with Node A, Node B and Node C. You have a cluster resource group "MyClusteredGroup" which contains various disk, IP, network name and service resources. Nodes A, B and C are all possible owners but Node B is set as the preferred owner and is currently the active node.

    The resource group fails over to Node C as Node B stops responding on the Public network due to a failed NIC. In the Resource group properties on the failback tab you have this set to immediate.

    You fix the NIC issue on Node B and bring it back up on the network. The resource group currently active on Node C will without warning immediately attempt to failback to Node B. Not a good idea if this is a Production SQL Server instance, so use caution when configuring preferred owners and failback 😉

    -----------------------------------------------------------------------------------------------------------

    "Ya can't make an omelette without breaking just a few eggs" 😉

  • Perry Whittle (9/14/2011)


    Possible owners are defined at the resource level and dictate which nodes in the Windows cluster are able to service this resource


    For instance, you have a 3 node cluster with Node A, Node B and Node C. You have a clustered disk resource "MyClusteredDisk", if you remove Node C from the possible owners of the clustered disk resource "MyClusteredDisk" then this disk will never be failed over to Node C.

    Preferred owners are defined at the resource group level and define the preferred node ownership within the Windows cluster


    For instance, you have a 3 node cluster with Node A, Node B and Node C. You have a cluster resource group "MyClusteredGroup" which contains various disk, IP, network name and service resources. Nodes A, B and C are all possible owners but Node B is set as the preferred owner and is currently the active node.

    The resource group fails over to Node C as Node B stops responding on the Public network due to a failed NIC. In the Resource group properties on the failback tab you have this set to immediate.

    You fix the NIC issue on Node B and bring it back up on the network. The resource group currently active on Node C will without warning immediately attempt to failback to Node B. Not a good idea if this is a Production SQL Server instance, so use caution when configuring preferred owners and failback 😉

    my understanding (and please correct me if I'm wrong) is that failback and preferred owners are not neccesaritly exclusive settings... for example, if I have already realized your suggestion above, and my cluster is still set to "prevent failback", I can still take advantage of "preferred owners" in order to say in effect "in the event of a failover of Node A, try Node B, only failover to Node C if both A and B are unavailble".... correct?

  • yes, prevent failback will do it exactly what it says. Set the policy to immediate or scheduled and the cluster group will comply, probably not what you want on a clustered production instance of sql server.

    This link details what would happen in a larger cluster consisting of 6 nodes

    -----------------------------------------------------------------------------------------------------------

    "Ya can't make an omelette without breaking just a few eggs" 😉

  • Thanks, Perry,

    If I have a sceanrio where I have 3 nodes namely A,B,C out of this A is active B is active an C is passive. If I have preferred nodes for A is in the order C, B, A. then if for some reasons If there is an issue with A it would failover to C first right?

    and for the same scenario above, on Node A one of the SQL Server Service possible owner order is as mentioned Node B, Node C, Node A

    if for some reasons if the NODE A goes down on which Nodes the service will move is it node C or node B looking at the order. thinking that only SQL Server Service has possible owner order with node B, Node c, Node A is it only SQL Server service would move to node B and the rest of them would be on NODE C?

    Scnearion 3 if I'm doing some patching and I don't want to un planned failover to happen and I want to remove the other nodes from failing over in preferred owners do I need to remove the nodes or from possible owner will I remove the other nodes or From both I will remove the other nodes?

    Thanks,

    Ravi

  • As far as I know, if the resource is managed on a Windows system, it can be managed using a workstation with privileged access, even if the application is running on a non-Windows operating system or based on a third-party cloud platform. For example, to manage an account in an Amazon Web Services subscription, an owner should use only a workstation with privileged access.

    https://theessayservice.org/

     

Viewing 8 posts - 1 through 7 (of 7 total)

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