File Formats                           cluster_definition.conf(4)



NAME
     cluster_definition.conf - nhinstall  configuration  file  to
     define the cluster

SYNOPSIS
     cluster_definition.conf

DESCRIPTION
     Configure the cluster_definition.conf file  to  define  your
     cluster for the nhinstall(1M) tool.

      The templates for the configuration files are contained  in
     the  /opt/SUNWcgha/config.standard  directory with .template
     extensions. Copy the configuration files to a  local  direc-
     tory on the installation server as follows:

     # mkdir <congif_dir>
     # export NHOME=/opt/SUNWcgha/config.standard
     # cd <config_dir>
     # cp $NHOME/env_installation.conf.template env_installation.conf
     # cp $NHOME/cluster_definition.conf.template cluster_definition.conf
     # cp $NHOME/addon.conf.template addon.conf

          Note:

          All the configuration files must be in the  same  local
          directory on the installation server.

EXTENDED DESCRIPTION
     Modify variables in the cluster_definition.conf file as fol-
     lows:

     CLUSTER_ID
           The cluster identifier used to assign IP addresses  to
           all  the  nodes  in the cluster.  This must be a value
           between 1 and 254. For example:

           CLUSTER_ID=254

     ARCH  The hardware architecture of the nodes in the cluster.
           Only sun4u is supported. For example:

           ARCH=sun4u

     MEN_INTERFACES
           The  network  interfaces  used  for  CGTP  on  master-
           eligible nodes.

           nic0 is the name of the first network interface.

           nic1 is the name of the second network interface.

           For example, the default is:

           MEN_INTERFACES= hme0 hme1

     NMEN_INTERFACES
           The network  interfaces  used  for  CGTP  on  diskless
           nodes.

           nic0 is the name of the first network interface.

           nic2 is the name of the second network interface.

           For example, the default is:

           NMEN_INTERFACES= eri0 eri1

     VENDOR_TYPE
           List of vendor types of diskless nodes to  boot.  This
           variable  may  have multiple entries. All entries will
           be used when configuring DHCP.

           The predefined vendor types are:

              o  SUNW.UltraSPARC-IIi-cEngine

              o  SUNW.UltraSPARC-IIi-Netract

              o  SUNW.UltraSPARCengine_CP-60

     VENDOR_TYPE enables you to add the vendor type  string  sent
     by  a  diskless  node when issuing a DHCP boot request. This
     enables you to include boards for diskless  nodes  that  are
     not part of the supported hardware.

           To add new vendor type:

           VENDOR_TYPE=<new_vendor_type>


     USE_SVM
           Install the the volume management feature of  Solaris.
           Options are YES or NO.

           If USE_SVM  is  set  to  YES,  the  volume  management
           feature  of  Solaris  is installed for managing repli-
           cated slices and their bitmap slices. If you are using
           FC-AL  disks on Netra 20 servers, you must set USE_SVM
           to YES.

           The volume management feature must be used  for  Netra
           20  hardware  because the disk scanning mechanism does
           not guarantee that  the  slot  position  of  the  disk
           provides  a  unique  and reproducible unit number (for
           example, when some disks are  plugged  on  FC-AL).  In
           this  case,  a slice with the attribute "replica" must
           be created. This slice is  reserved  for  storing  the
           metadevice database.

     NODE  The identifiers for each node. There is an  entry  for
           each node and each entry has the following format:

           NODE=<node-id> { <MAC-address> | - } [ - ] [<hostname> | - ] }


              o  <node-id>

                 The identifier of the node  used  to  assign  IP
                 addresses  to all the nodes in the cluster. This
                 must be a value between 3 and 254.

              o  <MAC-address>

                 The Ethernet address of the node.

                 For  master-eligible  nodes,  this  address   is
                 required  to  boot the master-eligible node from
                 the installation server.

                 For diskless nodes, this address is ignored.

              o  -

                 The hyphen indicates a node definition  reserved
                 for  future  use.

              o  <hostname>

                 <hostname> is optional.  If not specified, a  host
                 name  is  automatically  created  based  on  the
                 CLUSTER_ID definition and <node-id>.
           
           For example:

           NODE=10 08:00:20:f9:c5:54
           NODE=20 08:00:20:f9:a8:12
           NODE=30 -                               - node30
           NODE=40 -                               - node40


          Note:
          The first two entries must be the identifiers  for  the
          master-eligible  nodes. The nhinstall tool installs the
          product first on the first master-eligible node defined
          and then on the second master-eligible node.

     NIC0_POSTFIX, NIC1_POSTFIX, CGTP_POSTFIX
           Suffix string definitions. These  definitions  can  be
           set to change the suffix for each host name related to
           a specific network interface. By default the  suffixes
           are:


              o  NIC0_POSTFIX=""

                 Therefore, by default, the  host  name  for  the
                 NIC0 interface has no suffix.

              o  NIC1_POSTFIX="-nic1"

              o  CGTP_POSTFIX="-cgtp"

     If you specify all the suffixes, as shown below,  the  nhin-
     stall tool creates two host names in the /etc/hosts file-one
     host name with the suffix and one host name without the suf-
     fix with respect to the NIC0 interface.

     For example, if the suffixes are:


              o  NIC0_POSTFIX="-mynic0"

              o  NIC1_POSTFIX="-mynic1"

              o  CGTP_POSTFIX="-mycgtp"
           The resulting /etc/hosts file will include the follow-
           ing type of entry for each node:

                   o  10.128.1.10   MEN-C128-N10    MEN-C128-N10-
                      mynic0

                   o  10.128.1.11   MEN-C128-N11    MEN-C128-N11-
                      mynic0

                   o  10.128.1.12   NMEN-C128-N12   MEN-C128-N12-
                      mynic0

     SLICE
           SLICE defines the disk partitioning. There is an entry
           for  every  slice  on the disk, and each entry has the
           following format:

           SLICE=<name> <size> <mount-point> <bitmap> <option>

              <name>

                 The name of the slice.  For  example,  c0t0d0s0.
                 The  name  can  also be a metadevice name if you
                 have already installed Solaris  configured  with
                 the volume management feature. Note that slice 2
                 is reserved.

              <size>

                 The size of the slice in megabytes. For slice 7,
                 the  size can be set to free unless the slice is
                 replicated.

              <mount-point>

                 The name of the mount point of the slice. If the
                 slice  is  a  replication slice, the mount point
                 must be set to unnamed.

                 If the slice is reserved for metadevice database
                 storage, the mount point must be set to replica.
                 This is because one slice is required  with  the
                 replica attribute when USE_SVM is set to YES.

              <bitmap>

                 The name of the replicated slice. If  the  slice
                 is not replicated, <bitmap> must be set to - .

              <option>

                 The mount option.

                 For further details  about  <name>, <size>, <mount-point>, 
                 and <option>, see the Solaris JumpStartTM documentation.
           For example:

                SLICE=c0t0d0s0 2048 /               -        logging
                SLICE=c0t0d0s1 1024 swap            -        -
                SLICE=c0t0d0s3 2048 /export         c0t0d0s5 logging
                SLICE=c0t0d0s4 2048 /SUNWcgha/local c0t0d0s6 logging
                SLICE=c0t0d0s5 3    unnamed         -        -
                SLICE=c0t0d0s6 3    unnamed         -        -
                SLICE=c0t0d0s7 free /test1          -        logging


     EXPORTED
           The name of the directory to be created  and  exported
           on  the  master-eligible  node.  There is an entry for
           every directory to be exported.  Each  entry  has  the
           following format:

           EXPORTED=<name>

           Where  <name>  is  the  name  of  the  directory  to  be
           exported.

           The   nhinstall   tool   automatically   creates   the
           /SUNWcgha/local/export directory.

           If a diskless environment is required,  the  nhinstall
           tool automatically creates the /export directory.

           Do  not  manually  create  any  directories  that  are
           automatically created.

          Note:

          Exported directories must  be  on  a  replicated  slice
          because these directories must be accessible regardless
          of the master and vice-master roles.  Exporting a  non-
          replicated directory results in errors during a switch-
          over.

     MOUNTED
           The mount point on the master-eligible  node  that  is
           mounted  on  the  master-eligible  and diskless nodes.
           There is an entry for each mount point, and each entry
           has the following format:

           MOUNTED=<mounting-point> <remote-dir>


              <mounting-point>

                 The mount point on each node.

              <remote-dir>

                The directory name on the master-eligible node.

           The following directories are automatically defined by
           the nhinstall tool. Do not define them.

              o  /SUNWcgha/remote

                 The mount point for:

                 <master-name>:/SUNWcgha/local/export/data

              o  /SUNWcgha/services

                 The mount point for:
                 <master-name>:/SUNWcgha/local/export/services/ha_v1/opt

              o  /SUNWcgha/swdb

                 The mount point for:

                 <master-name>:/SUNWcgha/local/export/services

                Warning:

                For remote directories, define only child  direc-
                tories  that  were  previously exported, that is,
                directories on a replicated slice.

                You cannot use /export or its  child  directories
                as   automatically  exported  remote  directories
                because it will conflict  with  diskless  manage-
                ment.

EXAMPLES

     Example 1: Example of a standard configuration for a cluster
     with two master-eligible nodes and one diskless node

     #
     #  Cluster definition
     #
     CLUSTER_ID=150

     #
     # Architecture of Machines
     #
     ARCH=sun4u

     #
     # Network interfaces used for CGTP support on master-eligible nodes
     #

     MEN_INTERFACES=hme0 hme1
     #
     # Network interfaces used for CGTP support
     # on diskless nodes
     #
     NMEN_INTERFACES=hme0 hme1

     #
     # Node definition
     #
     NODE=10 08:00:20:f9:c5:54
     NODE=20 08:00:20:f9:a8:12
     NODE=30 08:00:20:f9:be:da

     #
     #  Definition of partitions for master-eligible nodes
     #
     SLICE=c0t0d0s0 2048 /               -        logging
     SLICE=c0t0d0s1 1024 swap            -        -
     SLICE=c0t0d0s3 2048 /export         c0t0d0s5 logging
     SLICE=c0t0d0s4 2048 /SUNWcgha/local c0t0d0s6
     SLICE=c0t0d0s5 3    unnamed         -        -
     SLICE=c0t0d0s6 3    unnamed         -        -
     SLICE=c0t0d0s7 free /test1          -        logging

     Example 2: Example of a configuration for a cluster with  no
     diskless nodes

     If you have only two master-eligible nodes in your  cluster,
     the  disk  layout  and  disk partitioning are different from
     those of a cluster with one diskless node or more.

     The order of the  slices  differs  from  the  order  in  the
     preceding example as follows:

        o  /SUNWcgha/local is slice 5 instead of slice 4.

        o  Replication occurs on slice 3  and  4  instead  of  on
           slices 5 and 6.

        o  There is no /export partition. Instead, there are  two
           user partitions-one replicated and one not replicated.

        o  The user directory is defined as exported on both par-
           titions,  and  the  mount points on the user directory
           are accessible from both nodes.

     #
     #  Cluster definition
     #
     CLUSTER_ID=150

     #
     # Architecture of Machines
     #
     ARCH=sun4u

     #
     # Network interfaces used for CGTP support on master-eligible nodes
     #
     MEN_INTERFACES=hme0 hme1

     #
     # Node definition
     #
     NODE=10 08:00:20:f9:c5:54
     NODE=20 08:00:20:f9:a8:12

     #
     #  Definition of partitions for master-eligible nodes
     #
     #  /user1 is a replicated partition, /test1 is not replicated
     #
     SLICE=c0t0d0s0 2048 /               -        logging
     SLICE=c0t0d0s1 1024 swap            -        -
     SLICE=c0t0d0s3 3    unnamed         -        -
     SLICE=c0t0d0s4 3    unnamed         -        -
     SLICE=c0t0d0s5 2048 /SUNWcgha/local c0t0d0s3 logging
     SLICE=c0t0d0s6 2048 /user1          c0t0d0s4 logging
     SLICE=c0t0d0s7 free /test1          -        logging

     #
     # definition of exported directories
     #
     EXPORTED=/user1/export

     #
     #  definition of mounted directories
     #  from master node to master (loop), vice-master and diskless
     MOUNTED=/user1_app /user1/export/app


     Example 3: Example of a configuration  for  a  cluster  with
     volume-managed replicated slices for FC-AL disks

     #
     #  Cluster definition
     #
     CLUSTER_ID=150

     #
     # Architecture of Machines
     #
     ARCH=sun4u

     #
     # selection of volume management for FC-AL disks
     #
     USE_SVM=YES

     #
     # Network interfaces used for CGTP support on master-eligible nodes
     #
     MEN_INTERFACES=hme0 hme1

     #
     # Network interfaces used for CGTP support
     # on diskless nodes
     #
     NMEN_INTERFACES=hme0 hme1

     #
     # Node definition
     #
     NODE=10 08:00:20:f9:c5:54
     NODE=20 08:00:20:f9:a8:12
     NODE=30 -

     #
     #  Definition of partitions for master-eligible nodes
     #
     SLICE=c0t0d0s0 2048 /               -        logging
     SLICE=c0t0d0s1 1024 swap            -        -
     SLICE=c0t0d0s3 2048 /export         c0t0d0s5 logging
     SLICE=c0t0d0s4 2048 /SUNWcgha/local c0t0d0s6 logging
     SLICE=c0t0d0s5 3    unnamed         -        -
     SLICE=c0t0d0s6 3    unnamed         -        -
     SLICE=c0t0d0s7 free replica         -        logging


ATTRIBUTES
     See attributes(5) for descriptions of the  following  attri-
     butes:

     ____________________________________________________________
    |       ATTRIBUTE TYPE        |       ATTRIBUTE VALUE       |
    |_____________________________|_____________________________|
    | Architecture                | SPARC                       |
    |_____________________________|_____________________________|
    | Availability                | SUNWnhins                   |
    |_____________________________|_____________________________|


SEE ALSO
     env_installation.conf(4),   addon.conf(4),    nhinstall(1M),
     nhadm(1M)

Netra HA Suite 2.0  Last change: May 31, 2002                  
