Top
PRIMECLUSTER  Installation and Administration Guide4.3

13.2.2 Creating Cluster Applications on the Global Zone

13.2.2.1 Building the Cluster

Refer to "Chapter 4 Preparation Prior to Building a Cluster" and "Chapter 5 Building a Cluster," and perform the initial setup of the cluster to the global zone.

Also, create the Gds resource, Fsystem resource, Gls resource, and cluster applications on the global zone as detailed below. For single-node cluster operations, creating these resources is not necessary. Perform the settings only for GDS or GLS as needed.

Creating temporal cluster applications is not necessary as well. Proceed to "13.2.3 Preparations for Migrating an Existing Solaris Environment to a Non-Global Zone."

13.2.2.2 Creating Gds Resources

This setup is necessary if sharing non-global zone images or if performing data takeover between non-global zones. Create on any one node a GDS shared class for each non-global zone and create to the shared classes a volume for non-global zone images and a volume for data takeover.

If using the system as a switchover file system, edit the /etc/vfstab.pcl for all nodes which comprise the cluster.

Example

If the GDS class for non-global zone zone-a is class0001, the volume for the non-global zone's root file system is volume 0001 (with the mount point being /zone-a-system, and the file system type being UFS), the volume for allocation of Oracle data being volume0002 (with the mount point being /zone-a-oracle, and the file system type being UFS), then write the /etc/vfstab.pcl as follows. Make sure the beginning of each line always starts with "#RMS#."

#RMS#/dev/sfdsk/class0001/dsk/volume0001  /dev/sfdsk/class0001/rdsk/volume0001 /zone-a-system ufs - no -
#RMS#/dev/sfdsk/class0001/dsk/volume0002 /dev/sfdsk/class0001/rdsk/volume0002 /zone-a-oracle ufs - no -

13.2.2.3 Creating Fsystem Resources

This setup is necessary if sharing GDS shared classes between non-global zones. If allocating non-global zones to a shared disk, also create Fsystem resources corresponding to the file systems into which the non-global zones are allocated. Perform this setup from any one node.

Note

To create non-global zones, set the mount point mode for the non-global zone root file system volume to 700 with all nodes. This setup is necessary for creating the non-global zone's root file system.

Example)

# chmod 700 /zone-a-system

13.2.2.4 Creating Gls Resources

If you want to notify the error to a cluster when it occurs with a network in a shared IP zone, set up a takeover virtual interface to the GLS and then register it as a Gls resource to the global zone user application.

Create the same number of takeover virtual interfaces as you want to notify a failure among virtual interfaces used in each shared IP zone.

Example

An example to meet the following two requirements is shown below:

  • Three shared IP zones (zone-a, zone-b, and zone-c) hold two LANs: management LAN (192.168.10.0/24) and a public LAN (192.168.20.0/24).

  • When a fault occurs to each LAN, the error is sent to the cluster.

In this case, you must perform the following procedure.

  1. Create a total of six takeover virtual interfaces of GLS.

    [Setup example of the takeover virtual interface]
    
    # /opt/FJSVhanet/usr/sbin/hanethvrsc print
     ifname     takeover-ipv4    takeover-ipv6
    +----------+----------------+--------------------------------------------------+
     sha11:65   192.168.10.11    -
     sha12:65   192.168.10.12    -
     sha13:65   192.168.10.13    -
     sha21:65   192.168.20.11    -
     sha22:65   192.168.20.12    -
     sha23:65   192.168.20.13    -
  2. Register Gls resources of management and public LANs for three cluster applications where each zone is registered.

    [Resource registration for the user application]
    
     User application(uApp1)
       -Cmdline resource(controls non-global zone(zone-a))
       -Gls resource(management LAN(sha11: 192.168.10.11 of 65))
       -Gls resource(public LAN(sha21: 192.168.20.11 of 65))
     User application(uApp2)
       -Cmdline resource(controls non-global zone(zone-b))
       -Gls resource(management LAN(sha12:192.168.10.12 of 65))
       -Gls resource(public LAN(sha22: 192.168.20.12 of 65))
     User application(uApp3)
       -Cmdline resource(controls non-global zone(zone-c))
       -Gls resource(management LAN(sha13: 192.168.10.13 of 65))
       -Gls resource(public LAN(sha23: 192.168.20.13 of 65))

Note

  • It is not possible to use the IP address for the takeover virtual interface registered as a Gls resource for shared IP zone transmission. Use for transmission the IP address allocated for the non-global zone by the Cmdline resource (shared IP control).

  • If specifying an interface coordinated with the NIC switching mode to the non-global zone network setup, it is necessary to change the standby interface's deactivation method from "Unplumb" to "Plumb" using the hanetparam command's -d option.

    # /opt/FJSVhanet/usr/sbin/hanetparam -d plumb
    
    # /opt/FJSVhanet/usr/sbin/hanetparam print
       Line monitor interval(w)           :5
       Line monitor message output (m)    :0
       Cluster failover (l)               :5
       Standby patrol interval(p)         :15
       Standby patrol message output(o)   :3
       NIC switching mode(d)              :Plumb
       Cluster failover in unnormality (c):OFF
       Line status message output (s)     :OFF

13.2.2.5 Creating Temporal Cluster Applications

On the global zone, create temporal cluster applications in the same number as that of the number of non-global zones.

For the temporal cluster applications corresponding to a given non-global zone, register a file system resource (Fsystem) which corresponds to each volume included in the Gds resource, Gls resource, and GDS class which in turn correspond to that non-global zone.

Start each of the temporal cluster applications created in this section and set their status to Online status in the operational system.