AIX > Administrator > High Availability

Follow These Steps for Configuration of PowerHA HyperSwap

Editor’s note: This is the second in a three-part series about PowerHA SystemMirror for AIX Enterprise Edition. Part one, “HyperSwap Helps Keep Storage From Becoming a Single Point of Failure” introduced the concept and benefits of HyperSwap in the PowerHA SystemMirror Enterprise Edition environment.

Having walked through the HyperSwap concept in part one, we’ll now explore how to configure HyperSwap in PowerHA, as well as what needs to be done at the AIX level to set it up. AIX provides support for HyperSwap by introducing various components such as the HyperSwap kernel extension, storage framework, etc. Illustrated in Figure 1, these HyperSwap-related components include:

  • Cluster Aware AIX (CAA) orchestrates cluster-wide actions.
  • PowerHA HyperSwap Kernel Extension works with CAA to coordinate actions with other nodes; analyzes the messages from the PowerHA framework and AIX Storage Framework (SFW); and determines proper swap actions.
  • AIX Storage Framework works as the AIX interface to storage and manages the status of storage; alerts the PowerHA HyperSwap Kernel Extension about I/O errors; gets swap decisions from the kernel extension; and sends orders to AIX PCM multiple path I/O.

Initial Disk Configuration

The first step is preparing the disks for use with HyperSwap. Before starting with disk configuration, the AIX administrator should check if AIX PCM driver is being used and the identified disks do not have any SCSI reservation policy set.

These settings can be checked while disks are active or in available state. To configure all disks that are part of storage subsystem to use the AIX PCM (AIX_AAPCM) driver, implement the manage_disk_drivers command. This requires a system reboot:

$ manage_disk_drivers –d device –o AIX_AAPCM

Use the devrsrv command to check if SCSI reservations are set on the identified hdisks:

$ devrsrv –c query –l hdiskX

If the command returns anything other than NO RESERVE, the administrator must change the disk reservation policy to no_reserve using the chdev command. This is required for all of the identified disks on all of the cluster nodes.

Once the reservation policy has been set to no_reserve, the next step is to create HyperSwap disks. One must prepare the disks pairs in the storage subsystems and AIX first, and then configure them in PowerHA. The steps are:

  • Select two disks, one from each storage subsystem to be mirrored for HyperSwap disks. In this example, these are hdiskA (from storageA) and hdiskB (from storageB). The command, located in /opt/ibm/dscli/bin can be used to display the disk attributes including storage system LSS ID.
  • Establish the connection path from hdiskA to hdiskB (from storageA using mkpprcpath).
  • Establish the connection path from hdiskB to hdiskA (from storageB using mkpprcpath).
  • Establish the volume pair of hdiskA and hdiskB in one direction (from storageA using mkpprc).
  • Enable HyperSwap for hdiskA on all nodes of the cluster (from all nodes using chdev). For example:
    $ chdev –a san_rep_cfg=migrate_disk –l hdiskX –U

    Repeat this step for all of the nodes.

Once these steps are successful, the secondary disk becomes unavailable. It’s changed to “defined” state and is ready to be used for HyperSwap integration with PowerHA.

AIX Tunable Related to HyperSwapM

After the disk configuration is done, you must set the tunable for HyperSwap, as shown in Table 1.

Table 1: HyperSwap-relevant AIX Tunable Settings

Name Component Value Description
dyntrk Protocol driver (fscsi) enabled This provides transparent I/O recovery if N_Port ID for a device is changed. This option is handled at the host bus adapter (HBA) level.
fc_err_recov Protocol driver (fscsi) fast_fail It allows you to detect Fibre Channel problems between the switch and the storage device.
hcheck_interval Disk driver (hdisk) 60 The time interval when a health check request is sent to the storage device. The default setting is 60 seconds.
rw_timeout Disk driver (hdisk) 30 (for DS8K) DS8K RW timeout value is 30 seconds.
timeout_policy Disk driver (hdisk) fail_path (for DS8K) This is the default value for DS8K.

Most of the values in Table 1 are the default settings for AIX V7.1 However, those using AIX 6.1 TL8 or higher might need to change the tunable values to the ones listed.

PowerHA Configuration

If all of the elements have been identified and set up correctly, the next set of steps should be fairly simple to execute. First, make sure necessary PowerHA SystemMirror Enterprise Edition filesets including “” are installed on all of the nodes.

Then fill in the CAA rhosts file (/etc/cluster/rhosts) with the IP address of all cluster nodes (the IP address mapping to hostname). This should be done on all of the nodes. Then restart the clcomd daemon.

Once this is done, configure the two-site multi-node “Stretched” cluster. Do so in AIX telnet session using: smitty sysmirror -> Cluster Nodes and Networks -> Multi Site Cluster Deployment -> Setup a Cluster, Nodes and Networks. Later choose the cluster repository disk and optionally specify a multicast IP address. Do this using: smitty sysmirror -> Cluster Nodes and Networks -> Multi Site Cluster Deployment -> Define Repository Disk and Cluster IP Address.

Once cluster definition is done and repository disk defined, verify and synchronize the cluster. This will ensure all other nodes of the cluster get the definition in their respective object data managers (ODMs). Later, create a volume group and file system(s) on the HyperSwap-capable disks on one node and import the definition on all of the other nodes. This make the volume group a shared VG.

Then, define storage and site association in PowerHA, using: smitty sysmirror-> Cluster Applications and Resources -> Resources -> Configure DS8800 Metro Mirror (In-Band) Resources -> Configure Storage Systems -> Add a Storage System. At this point, define the primary storage and associate it with the DC site. Repeat this step for the other storage as well. Next, set up the mirror group for the HyperSwap disks. Use the path: smitty sysmirror -> Cluster Applications and Resources -> Resources -> Configure DS8800 Metro Mirror (In-Band) Resources -> Configure Mirror Groups -> Add a Mirror Group. Choose either “user,” “system” or “cluster_repository” for the type of mirror group to be added.

After the mirror groups are added, create a resource group with its inter-site management policy set to either “Prefer Primary Site” or “Online on Either Site.” Then add the mirror group, volume group(s) and other resources such as service IP label(s), application controllers, etc.

Now the administrator should verify and synchronize the cluster configuration. This completes the configuration and integration of HyperSwap with PowerHA SystemMirror Enterprise Edition.

Kunal Langer is a technical solutions architect, Power Systems, AIX and PowerHA, with the Systems and Technology Group Lab-Based Services and Training.

comments powered by Disqus



2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store
IBMi News Sign Up Today! Past News Letters