-
CLUSTER_ID
-
The
cluster ID used to assign IP addresses to all the nodes in the cluster. This
parameter must be a value between 1 and 254. For example:
CLUSTER_ID=250
It is mandatory to define this parameter.
-
PASSWORD
-
The
password for the superuser set for all nodes in the cluster. Set this password
even if you manually install the Solaris operating system on the master-eligible nodes because
this password is required by the nhinstall tool to execute
the commands that create the diskless environment.
By default, the password is sunrules.
-
MEN_INTERFACES
-
The network interfaces used for Carrier Grade Transfer Protocol (CGTP) on
master-eligible nodes. The parameter has the following format:
MEN_INTERFACES=nic0 nic1
It is mandatory to define this parameter.
-
NMEN_INTERFACES
-
The network interfaces used for CGTP on diskless nodes. The parameter has
the following format:
NMEN_INTERFACES=nic0 nic1
It is mandatory to define this parameter if diskless nodes are defined.
-
LOGICAL_SLICE_SUPPORT
-
Install the volume management feature of the Solaris operating system. Options are YES and NO. The default is NO.
If LOGICAL_SLICE_SUPPORT is set to YES,
the volume management feature of the Solaris operating system is configured for managing
replicated partitions and their associated bitmap partitions.
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 (Fibre Channel-Arbitrated Loop).
If you are using FC-AL disks on Netra 20 servers, you must do one of
the following:
-
Set SOLARIS_INSTALL to YES
in the env_installation.conf file.
Set LOGICAL_SLICE_SUPPORT to YES.
Define a partition with the mount point set to replica
by using the SLICE parameter.
Complete the configuration and run the nhinstall(1m) tool.
-
Set SOLARIS_INSTALL to NO
in the env_installation.conf file.
Set LOGICAL_SLICE_SUPPORT to NO.
Install the Solaris operating system yourself on the master-eligible nodes configured
with the support of the Solaris Volume Manager (Solaris 9) or Solstice DiskSuite
4.2.1 (Solaris 8).
Create a disk partition with the attribute replica.
This partition is reserved for storing the metadevice database.
Complete the configuration and run the nhinstall(1m) tool.
-
SERIALIZE_SYNC
-
Determine how disk synchronization is performed. Values are YES
and NO. The default value is NO.
Synchronization is necessary following a switchover or the vice-master
booting, or when you request a full replication. In these circumstances, if SERIALIZE_SYNC is set to NO, Reliable NFS starts
the synchonization of all slices at the same time. If SERIALIZE_SYNC is set to YES, slices are synchronized one slice
at a time. This reduces the network and disk overhead but increases the time
it takes for the vice-master to synchronize with the master. During this time,
the vice-master is not eligible to take on the role of master.
-
USE_WDT
-
Install
and configure the Watchdog Timer packages. Options are YES
and NO. The default is NO.
Set the USE_WDT parameter to YES
only if you are using Netra servers with hardware watchdogs at the LOM level.
When this parameter is set to YES, the Watchdog Timer is
installed and configured. In this case, you must install LOM packages. You
can install these packages using the addon.conf file.
For more information, see the addon.conf(4) man page.
Set the USE_WDT parameter to NO
if you are using Netra servers with hardware watchdogs at the OPB level or
if you do not want to install the watchdog timer. OPB-level hardware watchdogs
are monitored by the server's software.
-
EXTERNAL_ACCESS
-
Creates a hostname and an external IP address for the node that has the master
role. This IP address is the same for either master-eligible node when that
node takes on the master role. When a node takes on the master role, this
IP address and hostname are assigned to the master node.
The EXTERNAL_ACCESS parameter has the following format:
EXTERNAL_ACCESS=hostname IP-address INTERFACE
-
hostname is the external host name
given to the master node.
-
IP-address is the external IP address
given with the master node.
-
NIC is the physical network interface
to which the IP address is to be associated.
-
VENDOR_TYPE
-
List of platform names of diskless nodes to boot. This parameter can have
multiple entries. All entries will be used when the nhinstall
tool configures the DHCP.
The VENDOR_TYPE parameter enables you to add a string
defining the vendor type 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 tested hardware.
The predefined platform names are:
-
SUNW.UltraSPARC-IIi-cEngine
-
SUNW.UltraSPARC-IIi-Netract
-
SUNW.UltraSPARCengine_CP-60
-
SUNW.UltraAX-i2
-
SUNW.NetraCT-410
-
SUNW.NetraCT-810
To add new vendor type:
VENDOR_TYPE=new_vendor_type
-
USE_CGTP
-
Install
the CGTP on the cluster. Options are YES and NO. The default is YES.
If USE_CGTP is set to NO, the
CGTP packages and patches are not installed. Only the first network interface
is considered in the definition of NODE. The CGTP_POSTFIX, if specified, is ignored. In this case, you configure
a single network link and your cluster network is not redundant.
-
PATCH_WITH_SMOSSERVICE
-
Install diskless patches with smosservice. Options
are YES and NO. The default is NO.
If PATCH_WITH_SMOSSERVICE is set to YES, the smosservice patch command installs the
patches listed in the addon.conf file and installed on
the diskless environment after the Solaris installation. This installation
corresponds to phase I or S as described
in the addon.conf file. The patchadd
command installs all patches applied during phase F. The patchadd command is also used to install any temporary patches,
that is patches with names prefixed T.
-
Do not use PATCH_WITH_SMOSSERVICE on the
Solaris 9 operating system.
-
You can supersede the patch installation method set with PATCH_WITH_SMOSSERVICE on a per patch basis. For more information,
see the addon.conf(4) man page.
-
IDE_SUPPORT
-
Configure the Sun StorEdge Network Data Replicator (SNDR) to support IDE disks.
Options are YES and NO. The default
is NO. For example:
IDE_SUPPORT=YES
-
RESTRICT_RHOSTS
-
Restrict the master-eligible nodes from connecting to each other remotely.
Options are YES and NO. The default
is NO.
If RESTRICT_RHOSTS is set to YES,
only the installation server can connect remotely to the master-eligible nodes
with rsh and rcp commands. The master-eligible
nodes cannot connect to each other remotely. Therefore, you will not be able
to run nhadm(1m) commands such as synccheck and syncgen on the master-eligible nodes.
If RESTRICT_RHOSTS is set to NO,
the master-eligible nodes can connect to each other remotely with the rsh and rcp commands. The installation server
can also connect to the master-eligible nodes remotely.
-
DISKLESS_BOOT_POLICY
-
Define the boot policy of the diskless nodes based on a boot policy.
Options are DHCP_STATIC, DHCP_DYNAMIC,
and DHCP_CLIENT_ID. The default is DHCP_DYNAMIC.
To choose the DHCP dynamic boot policy, specify DHCP_DYNAMIC. In this case, the IP addresses are attributed randomly to diskless
nodes from a pool of IP addresses.
To choose the DHCP static boot policy, specify DHCP_STATIC. In this case, the IP addresses are attributed according to the
Ethernet address of the diskless nodes. If DHCP_STATIC
is selected, the MAC0 and MAC1
attributes for diskless nodes must be set in the NODE parameter.
Also, in the NODE parameter, the NIC0
and NIC1 values can be set if the default values
specified by NMEN_INTERFACE need to be superseded.
To choose the DHCP client ID boot policy, specify DHCP_CLIENT_ID. In this case, the IP addresses are attributed to a diskless node
based on its client ID. This address assignment scheme is relevant only for
CompactPCI servers.
-
REPLICATED_DHCP_FILES
-
Select the location of the DHCP configuration files. Options are YES and NO. The default is YES.
If REPLICATED_DHCP_FILES is set to YES,
the DHCP configuration files are located on a replicated partition and are
shared by the master-eligible nodes. If REPLICATED_DHCP_FILES
is set to NO, the DHCP configuration files are duplicated
on a local partition of each master-eligible node.
-
CLUSTER_NETWORK
-
Define the class of IP addresses for your cluster network. The cluster network
can have IP addresses of any class. The parameter has the following format:
CLUSTER_NETWORK=netmask nic0-subnet nic1-subnet cgtp-subnet
-
netmask is the mask common to all
subnets configured in the /etc/netmasks file, for example, 255.255.0.0.
-
nic0-subnet is the subnet of the
first network interface, NIC0. This subnet is configured
in the /etc/netmasks file, for example, 172.15.0.0.
-
nic1-subnet is the subnet of the
second network interface, NIC1. This subnet is
configured in the /etc/netmasks file, for example, 172.16.0.0.
-
cgtp-subnet is the subnet of the
virtual network interface, cgtp0. This subnet is
configured in the /etc/netmasks file, for example, 172.17.0.0.
By default, class C IP addresses are used as follows:
CLUSTER_NETWORK=255.255.255.0 10.clusterid.1.0 10.clusterid.2.0 10.clusterid.3.0
Where clusterid is the value of the CLUSTER_ID parameter.
-
DISKLESS_TYPE
-
Define whether the diskless nodes are to be installed in a cluster or as standalone
nodes running only CGTP. Options are CLUSTER and STANDALONE. The default is CLUSTER.
If you specify the CLUSTER option, the diskless nodes
are installed with all the Foundation Services. The nodes are configured to be part
of the cluster.
If you specify the STANDALONE option, the diskless
nodes are installed with the CGTP only. The nodes are configured to run as
standalone nodes that do not run all the Foundation Services. The two master-eligible
nodes are installed with the Foundation Services so that the standalone diskless node
can be booted by the master node.
-
NODE
-
Define each
node. It is mandatory to define this parameter.
There is an entry for each node and each entry has the following format:
NODE=nodeid {MAC0|client-id|-} {MAC1|-} {name|-} {NIC0|-} {NIC1|-} public-name public-IP public-NIC
-
nodeid
The ID of the node used to define IP addresses for the node. This option
is mandatory.
-
MAC0 | client-id | -
The Ethernet address of the first network interface of the node. This
option is mandatory.
-
For master-eligible nodes, this address is required to boot
the master-eligible node from the installation server.
-
For a diskless node, whether this value is required depends
on the DHCP boot policy used:
For the DHCP static boot policy, when the DISKLESS_BOOT_POLICY is set to DHCP_STATIC, specify the Ethernet
address of the first network interface of the node.
For the DHCP dynamic boot policy, when the DISKLESS_BOOT_POLICY is set to DHCP_DYNAMIC, the MAC0 address is ignored and you must specify a hyphen (-).
For the DHCP client ID boot policy, when the DISKLESS_BOOT_POLICY is set to DHCP_CLIENT_ID, specify the client
ID as a string in double quotation marks. You can insert a hexadecimal value,
for example "/00999:88:05".
Note If you want to include the back slash character, you must include
it twice.
-
MAC1
The Ethernet address of the second network interface of the node. This
option is mandatory.
-
For master-eligible nodes, the MAC1
address is ignored and you must specify a hyphen (-).
-
For a diskless node, this value is required depending on the
DHCP boot policy used:
For the DHCP static boot policy, when the DISKLESS_BOOT_POLICY is set to DHCP_STATIC, specify the Ethernet
address of the second network interface of the node.
For the DHCP dynamic boot policy, when the DISKLESS_BOOT_POLICY is set to DHCP_DYNAMIC, the MAC1 address is ignored and you must specify a hyphen (-).
For the DHCP client ID boot policy, when the DISKLESS_BOOT_POLICY is set to DHCP_CLIENT_ID, the MAC1 address is ignored and you must specify a hyphen (-).
-
name
Name of the node. By default, the names are assigned as follows:
-
For a master-eligible node, MEN-Cclusterid-Nnodeid.
-
For a diskless node, NMEN-Cclusterid-Nnodeid.
Do not use underscores ("_") when naming a node. For more
information, see hosts(4).
-
NIC0
Name of the first network interface. This parameter can be set for a
diskless node with the DHCP static boot policy only. If you are using the
DHCP static boot policy and you want to use the default value, which is the
first value defined by the NMEN_INTERFACES parameter, specify
a hyphen (-).
For master-eligible nodes and for diskless nodes with other types of
DHCP boot policies, the NIC0 address is ignored.
In these cases, you must specify a hyphen (-).
-
NIC1
Name of the second network interface. This parameter can be set for
a diskless node with the DHCP static boot policy. If you are using the DHCP
static boot policy and you want to use the default value, which is the second
value defined by the NMEN_INTERFACES parameter, specify
a hyphen (-).
For master-eligible nodes and for diskless nodes with other types of
DHCP boot policies, the NIC0 address is ignored.
In these cases, you must specify a hyphen (-).
-
public-name
Name of the node on the public network different from the name defined
with the name parameter. If PUBLIC_NETWORK is not defined, the public-name is ignored.
-
public-ip
IP address of the node on the public network. If PUBLIC_NETWORK is not defined, the public-ip is ignored.
-
public-nic
Network interface for the node supporting the public network. This can
be either a physical network interface or an alias. If PUBLIC_NETWORK is not defined, the public-nic is ignored.
Example 1:
NODE=10 08:00:20:f9:c5:54 - node10
NODE=20 08:00:20:f9:a8:12 - node20
NODE=30 - - node30
NODE=40 - - node40
Example 2:
NODE=10 08:00:20:f9:c5:54 - node10 - - FSNode1 192.168.12.5 hme1:5
NODE=20 08:00:20:f9:a8:12 - node20 - - FSNode2 192.168.12.6 hme1:101
NODE=30 - - node30
NODE=40 - - node40
The first two entries must be the IDs for the master-eligible nodes.
The nhinstall tool first installs the product on the first
master-eligible node defined and then on the second master-eligible node.
The remaining entries define the diskless nodes.
To add diskless nodes to a cluster that is already running, add the
definitions for the new nodes by using the NODE parameter
and run the nhinstall command with the add
option. For information on the nhinstall command to add
diskless nodes to the cluster, see the nhinstall(1m) man page.
-
PUBLIC_NETWORK
-
Define the public network IP addresses and netmasks for the cluster. The parameter
has the following format:
PUBLIC_NETWORK=netmask subnet
Note The PUBLIC_NETWORK parameter also configures
the network interface of the installation server. Therefore, the SERVER_IP in the env_installation.conf file
is an IP address that is part of the same subnet
as defined for the PUBLIC_NETWORK.
-
DEFAULT_ROUTER_IP
-
Defines the IP address of the default router for the public network.
The DEFAULT_ROUTER_IP is set to an IP address.
DEFAULT_ROUTER_IP=IP address
The default is the public IP address of the installation server.
-
NIC0_POSTFIX, NIC1_POSTFIX, CGTP_POSTFIX
-
Defines
a suffix string for each host name. These definitions can be set to change
the suffix for each host name related to a specific network interface. By
default the suffixes are:
If you specify all the suffixes, as in the following example, the nhinstall tool creates two host names in the /etc/hosts file--one host name with the suffix and one host name without
the suffix with respect to the NIC0 interface.
For example, if the suffixes are:
NIC0_POSTFIX="-mynic0"
NIC1_POSTFIX="-mynic1"
CGTP_POSTFIX="-mycgtp"
The resulting /etc/hosts file will include the
following type of entry for each node:
10.250.1.10 netraMEN1 netraMEN1-mynic0
10.250.1.20 netraMEN2 netraMEN2-mynic0
10.250.1.30 netraDISKLESS1 netraDISKLESS1-mynic0
Note Do not use underscores ("_") within a suffix. For
more information, see hosts(4).
-
DIRECT_LINK
-
You can prevent the occurence of two master nodes in one cluster, by connecting
the serial ports of the two master-eligible nodes and defining the DIRECT_LINK parameter. By default, this parameter is not configured.
The DIRECT_LINK parameter has the following format:
DIRECT_LINK=MEN1-serial-device MEN2-serial-device speed [heartbeat-in-seconds]
-
MEN1-serial-device is the serial
port on the first master-eligible node to use to connect to the second master-eligible
node, for example, /dev/ttya
-
MEN2-serial-device is the serial
port on the second master-eligible node to use to connect to the second master-eligible
node, for example, /dev/ttya
-
speed is the serial line speed.
Valid values for speed are 38400, 57600, 76800,
and 115200.
-
heartbeat-in-seconds is the frequency
of the heartbeat checking the link between the two master-eligible nodes.
The default value for heartbeat is 20 seconds.
-
BITMAP_IN_MEMORY
-
Define where the scoreboard bitmaps of the shared partitions are stored.
If you choose the option YES, the scoreboard bitmaps
are configured to be stored in memory. In this case, changes are written to
the disk only when the node is shut down. This provides better performance
on the nodes. However, if both master-eligible nodes fail, the disks must
be resynchronized.
Alternatively, if you choose the option NO, the scoreboard
bitmaps are configured to write changes to the disk at each update.
For compatibility with previous releases, the default is NO:
BITMAP_IN_MEMORY=NO
-
NFS_USER_DIR_NOAC
-
Define the NFS noac option for remote mounted directories.
The noac option suppresses data and attributes caching.
If you choose YES, the noac option
is configured when mounting remote directories. In this case, all data is
retrieved from the master node disk. Data and attribute caching is suppressed.
Alternatively, if you choose NO, the noac option is not configured. In this case, data is cached on the local
node.
Use the noac option if the impact on performance
is acceptable.
For compatibility with previous releases, the default is YES, for example:
NFS_USER_DIR_NOAC=YES
-
SLICE_SYNC_TYPE
-
Define how the replicated partitions are synchronized. Options are:
-
FS
-
Only blocks
that contain data are replicated. Choose this option for faster synchronization.
-
RAW
-
All blocks
are replicated. Choose this option for slower synchronization.
The default option for SLICE_SYNC_TYPE
is FS.
SLICE_SYNC_TYPE=FS
-
CHECK_REPLICATED_SLICES
-
Define whether the replicated partition sanity check is activated. Options
are YES and NO. The default is NO.
If CHECK_REPLICATED_SLICES is set to YES, the sanity check is activated. If CHECK_REPLICATED_SLICES is set to NO, the sanity check is not activated.
-
MASTER_LOSS_DETECTION
-
Define if the absense of a master node in the cluster must be detected
by diskless nodes. Options are YES and NO.
The default is YES.
If MASTER_LOSS_DETECTION is set to YES,
the absense of a master node in the cluster is detected by the diskless nodes
and if such an absense occurs, the diskless nodes are rebooted. If MASTER_LOSS_DETECTION is NO, the absense of a
master node in the cluster is not detected by diskless nodes.
-
SLICE
-
Defines
the disk partitioning. There is an entry for every partition on the disk,
and each entry has the following format:
SLICE=name size mount-point bitmap option
-
name
The name of the disk partition. For example, c0t0d0s0.
The name can also be a metadevice name if you have already
installed the Solaris operating system configured with the volume management feature. However,
slice 2 is reserved.
-
size
The size of the partition in Mbytes. For slice 7, the size can be set
to free unless the partition is replicated.
-
mount-point
The name of the mount point of the partition. If the partition is replicated,
the mount point must be set to unnamed.
If the partition is reserved for metadevice database storage, the mount
point must be set to replica. This is because one partition
is required to have the replica attribute when LOGICAL_SLICE_SUPPORT is set to YES. See Example
3 in this man page.
-
bitmap
The name of the replicated partition. If the partition is not replicated,
specify a hyphen (-).
-
option
The mount option.
For further details about name, size, mount-point, and option, see the Solaris JumpStart documentation.
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
The following disk partitions are mandatory:
-
EXPORTED
-
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
name is the directory to be exported. However,
do not specify the following directories:
-
/SUNWcgha/local/export because the nhinstall tool automatically creates this directory
-
/export if a diskless environment is
required, the nhinstall tool automatically creates this
directory
Note Exported directories must be on a replicated partition because
these directories must be accessible regardless of the node that is master.
Exporting a non-replicated directory results in errors during a switchover.
-
MOUNTED
-
The mount
point on the master-eligible node of the directory that is mounted on
the master-eligible nodes and the diskless nodes. There is an entry for each
mount point, and each entry has the following format:
MOUNTED=mounting-point remote-dir
The following directories are automatically created by the nhinstall tool. Do not define them.
-
The /SUNWcgha/remote directory, which
is the mount point for the men_name:/SUNWcgha/local/export/data directory.
-
The /SUNWcgha/services directory, which
is the mount point for the men_name:/SUNWcgha/local/export/services/ha_v1/opt.
-
The /SUNWcgha/swdb directory, which is
the mount point for the men_name:/SUNWcgha/local/export/services directory.
Where men_name is the host name of the master-eligible
nodes.
For remote directories, define only child directories that
were previously exported, that is, directories on a replicated partition.
Do not define /export
or its subdirectories because it will conflict with the management of diskless
nodes.
-
DATA_MGT_POLICY
-
Define how the cluster behaves when the vice-master node starts up while the
master node is down. Options are:
-
INTEGRITY
-
The
vice-master waits for the old master to rejoin the cluster before it takes
the master role. This ensures that the cluster uses the most up-to-date data.
-
AVAILABILITY
-
The vice-master does not wait for the old master to rejoin the cluster before
it takes the master role. Data written to the master while the vice-master
is down is lost.
-
ADAPTABILITY
-
The vice-master checks the disk synchronization state. If the state is not
synchronized, that is the state returned by nhcmmstat is synchro:NEEDED, the vice-master waits for the old master to rejoin
the cluster. If the state is synchronized, that is the state returned by nhcmmstat is synchro:READY, the vice-master is
elected as the new master.
The default value for DATA_MGT_POLICY
is INTEGRITY.
DATA_MGT_POLICY=INTEGRITY
-
SYNC_FLAG
-
Delays
disk synchronization at startup. Options are YES and NO. The default is YES.
If SYNC_FLAG is set to NO, you
delay the start of disk synchronization until you use the nhenablesync command. For more information, see the nhenablesync(1m) man page.
-
MEN_OS_REFERENCE
-
Defines the profile of the operating system to be used. This parameter is
used when the automatic release detection based on the /etc/release file does not enable you to determine the correct operating system
version where the Solaris distribution has been updated but there is no change
to the Solaris release identification information. Current options are:
-
S8U7
-
Solaris
8 2/02 s28s_u7wos_08a SPARC
-
S8U7_108528-21
-
Solaris 8 2/02 s28_uwos_08a SPARC -- patched version of the Solaris operating system
which is reserved for use with Netra CT 820 hardware
-
S8PSR3
-
Solaris
8 HW 7/03 s28s_hw3wos_05a SPARC
-
S9HWPL3
-
Solaris
9 s9_58shwp13 SPARC
-
S9U1
-
Solaris
9 9/02 s9s_u1wos_08b SPARC
There is no default value for MEN_OS_REFERENCE. The option is automatically detected based on the contents of
the /etc/release file.
-
DISKLESS_OS_REFERENCE
-
Defines the profile of the Solaris operating system used in the diskless node environment.
This parameter is used when automatic release detection based on the contents
of the /etc/release file does not permit nhinstall to determine the correct version of the operating system. This
can occur when the Solaris distribution has been updated but the release identification
information has not changed. Options are:
-
S8U7
-
Solaris
8 2/02 s28s_u7wos_08a SPARC
-
S8U7_108528-21
-
Solaris 8 2/02 s28_uwos_08a SPARC -- patched version of the Solaris operating system
which is reserved for use with Netra CT 820 hardware
-
S8PSR3
-
Solaris
8 HW 7/03 s28s_hw3wos_05a SPARC
-
S9HWPL3
-
Solaris
9 s9_58shwp13 SPARC
-
S9U1
-
Solaris
9 9/02 s9s_u1wos_08b SPARC
There is no default value for MEN_OS_REFERENCE. The option is automatically detected based on the contents of
the /etc/release file.