Configuring split and merge policies

You can use the SMIT interface to configure split and merge policies.

The possible split policy and merge policy configuration depends on the version of the AIX® operating system running in your environment.

The following table displays the possible split and merge policy combinations for different cluster types in AIX Version 7.2.0, or earlier:
Table 1. AIX Version 7.2.0, or earlier, split and merge policy combinations
Cluster type Split policy option Merge policy option
Standard cluster Not supported Majority
Stretched cluster None Majority
Tie breaker Tie breaker
Linked cluster None Majority
Tie breaker Tie breaker
Manual Manual

In AIX Version 7.1.5, or later, and AIX Version 7.2.1, or later, all standard, stretched, and linked clusters support the same split policy and merge policy combinations.

The following table displays the possible split and merge policy combinations in AIX Version 7.2.1, or later:
Table 2. AIX Version 7.2.1, or later, split and merge policy combinations
Split policy options Merge policy options
None Majority
None None
Tie breaker Tie breaker
Manual Manual
Cloud Cloud
To configure a split and merge policy in PowerHA® SystemMirror®, complete the following steps:
  1. From the command line, enter smit sysmirror.
  2. In the SMIT interface, select Custom Cluster Configuration > Cluster Nodes and Networks > Initial Cluster Setup (Custom) > Configure Cluster Split and Merge Policy > Split and Merge Management Policy, and press Enter.
  3. Complete the following fields, and press Enter.
    Note: Some of the following fields are displayed only when certain options are specified in the Split Handling Policy and Merge Handling Policy fields.
    Table 3. Split and Merge Management Policy fields
    Field Description
    Split Handling Policy You can select the following options for this field:
    None
    This option is the default setting. If you select this option, the partitions operate independently of each other after the cluster split occurs. If you select this option, the Merge Handling Policy field is set to Majority.
    Tie Breaker
    You can select the following options for the tie breaker:
    Disk
    Select this option to specify a disk as the tie breaker. When the cluster split occurs, one site wins the SCSI reservation on the tie breaker disk. The site that losses the SCSI reservation uses the recovery action that is specified in the policy setting. If you select this option, the Merge Handling Policy field is set to Tie breaker.
    NFS
    Select this option to specify an NFS file as the tie breaker. During the cluster split, a predefined NFS file is used to decide the winning partition. The partition that losses the NFS file reservation uses the recovery action that is specified in the policy setting. If you select this option, the Merge Handling Policy field is set to NFS.
    Cloud

    Select this option when the nodes are configured for cloud communication. When a cluster split event occurs, each partition attempts to acquire the lock by uploading a file to the configured Cloud service. The partition that successfully uploads the file to the configured Cloud service continues to function. The partition that cannot upload the file to the configured Cloud service is rebooted or the cluster services are restarted if it is the specified action in the policy setting.

    If you select this option, the Merge handling policy field is set to Cloud.

    Before you configure the split and merge policies as Cloud, you must enable Cloud communication on all nodes of the cluster and you must have access to create a Cloud Storage bucket for the Cloud service. During the verification and synchronization process, PowerHA SystemMirror verifies the communication to the Cloud and displays a warning for any failure in the cloud communication when the Split handling policy or Merge handling policy field is configured as Cloud. Also, you must install Python and Boto3 Software Development Kit (SDK) on all nodes of the cluster. For more information about the Boto3 Software Development Kit (SDK), see the Boto3 SDK Documentation website.

    Manual
    Select this option to wait for manual intervention when a cluster split occurs. PowerHA SystemMirror does not perform any actions on the cluster until you specify how to recover from the cluster split. If you select this option, the Merge handling policy field is set to Manual.
    Merge Handling Policy Do not change the options that are specified in this field. The options that are specified in this field are determined by the options you selected in the Split Handling Policy panel.
    Split and Merge Action Plan You can select the following options for this field:
    Reboot
    Select this option to reboot nodes on the losing partition when a cluster split event occurs.
    Disable Applications Auto-Start and Reboot
    Select this option to reboot nodes on the losing partition when a cluster split event occurs. If you select this option, the resource groups are not brought online automatically after the system reboots.
    Note: This option is available only if your environment is running AIX Version 7.2.1, or later.
    Disable Cluster Services Auto-Start and Reboot
    Select this option to reboot nodes on the losing partition when a cluster split event occurs. If you select this option, Cluster Aware AIX (CAA) is not started. The resource groups are not brought online automatically. After the cluster split event is resolved, in SMIT, you must select Problem Determination Tools > Start CAA on Merged Node to restore the cluster.

    If you specified None in the Merge Handling Policy and Split Handling Policy fields, the action plan is not implemented and a reboot does not occur after the cluster split event.

    Note: This option is available only if your environment is running AIX Version 7.2.1, or later.
    Select tie breaker This field is available if you select Tie Breaker - Disk in both the Split Handling Policy and the Merge Handling Policy fields.

    Select an iSCSI disk or a SCSI disk that you want to use as the tie breaker disk.
    start of changeSelect Site Priorityend of change This field is available if you select Tie Breaker - Disk in both the Split Handling Policy and the Merge Handling Policy fields.

    Select any cluster site in PowerHA SystemMirror that enables the tie breaker win during cluster split operation.

    Note: Site Priority configuration is functional only with Reliable Scalable Cluster Technology (RSCT) Version 3.3.2.0, or later.
    start of changeTime Delay for Lower Priority Siteend of change This field is available if you select Tie Breaker - Disk in both the Split Handling Policy and the Merge Handling Policy fields.

    To reserve the tie breaker, specify the time in seconds to delay the lower priority site. You can enter a value in the range 0 – 300 seconds. The default value is 30 seconds only when Site Priority option is selected.
    NFS Export Server This field is available if you specify Tie Breaker - NFS in both the Split Handling Policy and the Merge Handling Policy fields.
    Specify the fully qualified domain name of the NFS server that is used for the NFS tie-breaker. The NFS server must be accessible from each node in the cluster by using the NFS server IP address.
    Note: To export NFS from a Linux® operating system, you must disable the dir_index option by running the tune2fs -O ^dir_index <filesystem> command. Where, <filesystem> is the NFS directory.
    Local Mount Directory This field is available if you specify Tie Breaker - NFS in both the Split Handling Policy and the Merge Handling Policy fields.

    Specify the absolute path of the NFS mount point that is used for the NFS tie-breaker. The NFS mount point must be mounted on all nodes in the cluster.

    Note: The default NFS mount options are vers=4,fg,soft,retry=1,timeo=10. Modifying the default values might lead to failure in acquiring the NFS lock.
    NFS Export Directory This field is available if you specify Tie Breaker - NFS in both the Split Handling Policy and the Merge Handling Policy fields.

    Specify the absolute path of the NFSv4 exported directory that is used for the NFS tie-breaker. The NFS exported directory must be accessible from all nodes in the cluster that use NFSv4.

    You must verify that the following services are active in the NFS server:
    • biod
    • nfsd
    • nfsgryd
    • portmap
    • rpc.lockd
    • rpc.mountd
    • rpc.statd
    • TCP
    You must verify that the following services are active in the NFS client on all cluster nodes:
    • biod
    • nfsd
    • rpc.mountd
    • rpc.statd
    • TCP
    • portmap
    Notify Method This field is available if you specify Manual in both the Split Handling Policy and the Merge Handling Policy fields.

    Specify the method that informs you to select which partition remains online after a cluster split or merge event. The method is specified as a path name that is followed by optional parameters. When this method is called, the last parameter is either split or merge to indicate the type of cluster event.

    Notify Interval This field is available if you specify Manual in both the Split Handling Policy and the Merge Handling Policy fields.

    Enter the seconds between notification messages that are sent asking you to select which partition remains online after a cluster split or merge event. You can enter a value in the range 10 - 3600.

    Maximum Notifications This field is available if you specify Manual in both the Split Handling Policy and the Merge Handling Policy fields.

    Enter the number of times you want a notification message to be sent to select which partition remains online after a cluster split or merge event. You can enter a value in the range 3 - 1000. You must enter a value for this field when the surviving (winning) site is specified.
    Default Surviving Site This field is available for only linked clusters and if you specified Manual in both the Split Handling Policy and the Merge Handling Policy fields.

    Enter the site that remains operational if you do not manually specify the surviving site. The other site that you do not specify in this field uses the action that is specified in the Split and Merge Action Plan field.

    Apply to Storage Replication Recovery This field is available for only linked clusters and if you have specified Manual in both the Split Handling Policy and the Merge Handling Policy fields.

    This field determines if the manual response when a cluster split event occurs also applies to storage replication recovery mechanisms. If you select Yes, the partition that was selected to remain online after a cluster split event proceeds with the takeover of the storage replication recovery. You cannot specify this option if you are using IBM® XIV® Storage Systems or IBM System Storage® DS8000®, or later.
  4. Verify that all fields are correct and press Enter.
  5. Verify and synchronize the changes across the cluster.
    Notes:
    • Split and merge policy are effective only in an active cluster.
    • The Deadman mode sets to an Assert mode when the cluster services start.