 |
|
|
NAME
nhfs.conf - Foundation Services configuration
file
SYNOPSIS
/etc/opt/SUNWcgha/nhfs.conf
The nhfs.conf file contains configuration information
for the Foundation Services such as the Cluster Membership Manager, Reliable NFS, and
the Node State Manager. This file also provides cluster addressing and interface
configuration information.
Configure this file on each node if you plan to install the Foundation Services
manually. To manually configure the parameters in the nhfs.conf file, uncomment the parameters, that is, delete the comment mark
(#) at the beginning of the line, and modify the value of the parameter. A
description of each parameter is provided in the following sections.
When you have manually install the Foundation Services packages, copy the template
file /etc/opt/SUNWcgha/nhfs.conf.template to the default
location /etc/opt/SUNWcgha as nhfs.conf
on each peer node. For each file, make the necessary modifications in a text
editor.
Do not re-edit the nhfs.conf file when the cluster
is running.
The nhfs.conf file format is ASCII. Parameters
consist of a keyword followed by an equals (=) sign followed by the parameter
value, of the form:
Keyword=Value
The following Keyword and Value parameters are supported.
-
Cluster.DataManagementPolicy
-
Define how the cluster behaves when the vice-master node starts
up while the master node is down. You can select one of three data management
policies. Values are Integrity, Availability,
and Adaptability. The default value is Integrity.
Cluster.DataManagementPolicy=Integrity | Availability | Adaptability
-
Integrity
-
If
there is a vice-master but no master in the cluster, the vice-master waits
for the old master to rejoin the cluster before it takes the master role.
This ensures that the cluster has the most up-to-date data. Choosing this
value might reduce the availability of the cluster, but it prioritizes data
integrity.
-
Availability
-
If there is a vice-master in the cluster but no master, the vice-master does
not wait for the old master to rejoin the cluster before taking on the master
role. The previous master is still off-line. Any data that is written to the
new master while there is no vice-master will be lost. When the old master
node comes back on line as the vice-master node, a full synchronization occurs
between the two master-eligible nodes.
-
Adaptability
-
If there is a vice-master in the cluster but no master, 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 master to come up. This is equivalent
to the Integrity data management policy. If the state is
synchronized, that is the state returned by nhcmmstat is synchro:READY, the vice-master is elected the new master. This is
equivalent to the Availability data management policy.
-
Cluster.Master.ID
-
Specify the host part of the master node's floating IP address in decimal
form. The floating address triplet for the node with the master role is calculated
from the local network interface addresses, the netmask, and the value of
this parameter. These addresses are IPv4 address.
The default value is 1.
Cluster.Master.ID=1
If the Cluster.Master.ID is 1
and the cluster network is set up as follows:
The floating address triplet is as follows:
-
NIC0 floating address: 192.200.168.1
-
NIC1 floating address: 10.250.2.1
-
cgtp0 floating address: 192.200.175.1
This address is calculated similarly if you have IP addresses of any
other class.
-
CMM.MasterLoss.Detection
-
Determine if the absense of a master node in the cluster must be detected
by diskless and dataless nodes. Values are True and False. The default is False.
If CMM.MasterLoss.Detection is set to False, the diskless and dataless nodes are rebooted when they detect
that there is no master node in the cluster for a period of more then 4-5
minutes. If CMM.MasterLoss.Detection is set to True, the diskless and dataless nodes do not reboot if there is
no master node in the cluster. However, the diskless and dataless nodes cannot
access directories that would be exported by the master node if such a node
existed. You must ensure that diskless and dataless nodes can adapt to this
situation.
-
Node.Domainid
-
Specify the domainid of the cluster. You must modify
this parameter. There is no default value.
Node.Domainid=250
-
Node.NodeId
-
This parameter specifies the node ID of the current node. There is no default
value.
-
Node.NIC0
-
The
name of the first network interface, NIC0, used
for CGTP. The default value is hme0.
Node.NIC0=hme0
This parameter could be a logical network interface, for example, hme0:2.
If you have not installed the CGTP patches and packages and you want
to configure a single network link for your cluster, configure this parameter.
-
Node.NIC1
-
The
name of the second network interface, NIC1, used
for CGTP. The default value is hme1.
Node.NIC1=hme1
This parameter could be a logical network interface, for example, hme1:2.
If you have not installed the CGTP patches and packages and you want
to configure a single network link for your cluster, do not configure this
parameter.
-
Node.NICCGTP
-
The virtual interface used by CGTP. The default value is cgtp0.
Node.NICCGTP=cgtp0
If you have not installed the CGTP patches and packages and you want
to configure a single network link for your cluster, do not configure this
parameter.
-
Node.UseCGTP
-
Specify whether the CGTP is to be used or not. Values are True
or False. The default value is True.
Node.UseCGTP=True
If you have not installed the CGTP patches and packages and you want
to configure a single network link for your cluster, set this parameter to False.
-
Node.Type
-
Specify
the type of node. Defining this parameter is mandatory. When you create the nhfs.conf file on each peer node, specify the type of node in this
parameter. The Node.Type parameter can have one of the
following values:
-
Diskfull--A master-eligible node
-
Dataless--A dataless node
-
Diskless--A diskless node
There is no default value. For more information on types of nodes, see
the Netra High Availability Suite Foundation Services 2.1 6/03 Overview.
-
Node.RNFS.Installed
-
Specify whether Reliable NFS is installed on a node. Values are True or False. The default value is False.
Node.RNFS.Installed=True
Both master-eligible nodes must have the same value for this parameter.
To prevent a split brain situation, you can connect the serial ports
of the master-eligible nodes and configure the following parameters. Split
brain is a situation where the network fails and two master nodes are elected
in the same cluster. Configure the following parameters if you have connected
the serial ports of the master-eligible nodes.
-
Node.Direct-Link.Serial.Device
-
Specify the serial device, that is, the system's serial ports.
There is no default value.
Node.Direct-Link.serial.Device=/dev/term/b
-
Node.Direct-Link.Serial.Speed
-
Specify the serial line speed. Valid values are 38400, 57600,
76800, or 115200. The higher the value, the better the link.
There is no default value.
Node.Direct-Link.serial.Speed=115200
-
Cluster.Direct-Link.Backend
-
This parameter enables the direct link. If this parameter
is not present in the nhfs.conf file, the direct link will
not be used even if you have connected the serial ports of the master-eligible
nodes.
The only value accepted by this parameter is serial:
Cluster.Direct-Link.Backend=serial
-
Cluster.Direct-Link.Heartbeat
-
Specify the number of seconds between two checks to detect a failure.
Therefore, if one master-eligible node receives no heartbeat during the specified
period of seconds, the node is alerted that there may be a problem.
There is no default value. For example, specify an interval of 20 seconds
as follows:
Cluster.Direct-Link.Heartbeat=20
-
CMM.IsEligible
-
Specify whether the node is master eligible. Values are True
or False. The default value is False.
CMM.IsEligible=True
-
CMM.LocalConfig.Dir
-
Specify the directory where the configuration file, cluster_nodes_table, is located on each master-eligible node. There is no default
value.
CMM.LocalConfig.Dir=/etc/opt/SUNWcgha
-
CMM.Miniconfig.Dir
-
Specify the directory where the configuration file, target.conf, is located on each master-eligible node. There is no default
value.
CMM.Miniconfig.Dir=/etc/opt/SUNWcgha
-
CMM.Miniconfig.File
-
Specify the name of the local configuration file, target.conf. There is no default value.
CMM.Miniconfig.File=target.conf
For more information on the CMM parameters, see the nhcmmd(1m)
man page.
-
RNFS.StatdAlternatePath
-
Specify the alternate statd repository. For information
about statd, see the statd(1m) man page.
You must modify this parameter. There is no default value.
RNFS.StatdAlternatePath=directory-path
where directory-path is the path to the statd directory. This directory must be on a replicated disk partition.
RNFS.StatdAlternatePath=/SUNWcgha/local
-
RNFS.Slice
-
Define
the disk partitions that must be replicated by Reliable NFS. You must modify
this parameter. The format for this parameter is:
RNFS.Slice.x=local-disk-partition local-bitmap-partition \
remote-disk-partition remote-bitmap-partition mount-flag
-
x is an integer (0, 1, 2) that
distinguishes each partition to be replicated.
-
local-disk-partition and remote-disk-partition are the paths to the raw disk partitions
on the local and remote nodes.
-
local-bitmap-partition and remote-bitmap-partition are the paths to the scoreboard bitmaps
associated to local-disk-partition and remote-disk-partition respectively.
-
mount-flag is 0, 1 or 2:
- 0
-
Do not mount the partition.
- 1
-
Mandatory mount. If the partition
is not mountable, stop with error.
- 2
-
Best effort mount. Try to mount
the partition. If this is not possible, log an error and continue.
There is no default value.
RNFS.Slice.0=/dev/rdsk/c0t0d0s3 /dev/rdsk/c0t0d0s5 \
/dev/rdsk/c0t0d0s3 /dev/rdsk/c0t0d0s5 1
-
RNFS.Share
-
Describes
the file systems to be shared. You must modify this parameter.
-
This parameter uses the same syntax as the share command. For more information, see the share(1m) man page.
-
Each share command must be on its own line. The first string
of an RNFS.Share line must be either share
or /usr/sbin/share.
-
Each RNFS.Share line must contain one share
command line per partition to be exported by NFS.
-
Each RNFS.Share.x
represents a partition to be shared and x is an
integer (0, 1, 2) that distinguishes each partition to be replicated.
To ensure that the CMM behaves correctly, you must:
-
Grant superuser access for all the master-eligible nodes on
the exported file system where the cluster_nodes_table
file resides.
-
Use the addresses for the CGTP interface for the master-eligible
nodes. If CGTP is not installed, use the NIC0 addresses.
There is no default value. In the following example, cgtp6 and cgtp7 represent the static CGTP addresses
of the master-eligible nodes.
RNFS.Share.0=share -F nfs -o rw,root=cgtp6:cgtp7 \
-d "SUNWcgha" /SUNWcgha/local/export
RNFS.Share.1=share -F nfs -o rw -d "diskless1" -o \
rw,root=diskless1:diskless1-b:cgtp-diskless1 \
/export/root/netraDISKLESS1
If you have diskless nodes in the cluster, add the cgtp0
address of the diskless nodes.
RNFS.Share.0=share -F nfs \
-o rw,root=cgtp6:cgtp7:cgtp8 \
-d "SUNWcgha" /SUNWcgha/local/export
-
RNFS.CheckReplicatedSlices
-
Check the status of replicated slices by continuously scanning them.
This property is disabled by default. To enable this property, set RNFS.CheckReplicatedSlices to True:
RNFS.CheckReplicatedSlices=True
This check is only performed when the cluster is in the synchro:READY state. To determine the synchronization state of the cluster, run
the nhcmmstat command. If you enable this property and
the cluster is not in the synchro:READY state the following
action occurs:
-
An error message is displayed on the master node.
-
The cluster is put in the synchro:NEEDED
state to prevent a switchover occurring. If a switchover occurs, the master
might not have access to the most up to date data.
-
RNFS.CheckReplicatedSlicesInterval
-
Set the time between two successive reads of the replicated
slices. Values are a number of milliseconds. The default value is 10 milliseconds.
The RNFS.CheckReplicatedSlicesInterval property is ignored
if the RNFS.CheckReplicatedSlices property is not set to True.
RNFS.CheckReplicatedSlicesInterval=time in milliseconds
-
RNFS.EnableSync
-
Determine when disk synchronization occurs. Values are True
and False. The default is True.
If RNFS.EnableSync is set to True,
disk synchronization is triggered at startup.
If RNFS.EnableSync is set to False,
you delay the start of disk synchronization until you use the nhenablesync command. For more information on this command, see the nhenablesync(1m)
man page.
RNFS.EnableSync=False
-
RNFS.SyncType
-
Specify the method used for synchronization between slices on master-eligible
nodes. Values are FS or RAW. The default
value is FS.
If RNFS.SyncType is set to FS,
this feature is enabled. The time taken for a full slice synchronization is
reduced because only the blocks used by the slice's file-system are replicated.
If RNFS.SyncType is set to RAW, this
property is disabled.
Both master-eligible nodes must have the same value for the RNFS.SyncType parameter. If you change the value of the RNFS.SyncType property, reboot the master-eligible nodes one at
a time. For information on rebooting cluster nodes, see "Shutting Down
and Restarting a Cluster" in the Netra High Availability Suite Foundation Services 2.1 6/03 Cluster Administration Guide. The RNFS.SyncType property is only valid for master-eligible nodes.
-
RNFS.SerializeSync
-
Determine how disk synchronization is performed. Values are TRUE and FALSE. The default value is FALSE.
Synchronization is necessary following a switchover or the vice-master
booting, or when you request a full replication. In these circumstances, if RNFS.SerializeSync is set to FALSE, Reliable
NFS starts the synchonization of all slices at the same time. If RNFS.SerializeSync is set to TRUE, 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.
For more information on the Reliable NFS parameters,
see the nhcrfsd(1m) man page.
Note Some parameters in this file require a single value while others
require multiple values.
RNFS.StatdAlternatePath=/SUNWcgha/local
RNFS.Slice.0=/dev/rdsk/c0t0d0s3 /dev/rdsk/c0t0d0s5 /dev/rdsk/c0t0d0s3 \
/dev/rdsk/c0t0d0s5 1
RNFS.Slice.1=/dev/rdsk/c0t0d0s4 /dev/rdsk/c0t0d0s6 /dev/rdsk/c0t0d0s4 \
/dev/rdsk/c0t0d0s6 1
RNFS.Share.0=share -F nfs -o rw,root=b14-ct400-10-cgtp:b14-ct400-20-cgtp \
-d "SUNWcgha" /SUNWcgha/local/export
RNFS.Share.1=share -F nfs -o rw -d "Export" /export
Configure these parameters to specify the external floating address
assigned to the master node. The following parameters must be configured when
you use the Node State Manager.
-
NSM.Exec.MasterDir
-
The directory containing the scripts for the master node. There is no
default value.
NSM.Exec.MasterDir=/opt/SUNWcgha/actions/master
-
NSM.Exec.ViceMasterDir
-
The directory containing the scripts for the vice-master node. There
is no default value.
NSM.Exec.ViceMasterDir=/opt/SUNWcgha/actions/vicemaster
-
NSM.Log.MasterDir
-
The log file directory for the master node. There is no default value.
NSM.Log.MasterDir=/var/run/SUNWcgha/actions/master
-
NSM.Log.ViceMasterDir
-
The log file directory for the vice-master node. There is no default
value.
NSM.Log.ViceMasterDir=/var/run/SUNWcgha/actions/vicemaster
-
NSM.External.Master.Address
-
The external floating address that is always assigned to the
master node.
NSM.External.Master.Address=IP-address/netmask-size
The netmask-size value is optional. There
is no default value.
NSM.External.Master.Address=192.168.12.39
-
NSM.External.Master.Broadcast
-
The broadcast address. This parameter is optional. If not
configured, the broadcast address is automatically determined based on the
IP address and netmask. There is no default value.
NSM.External.Master.Broadcast=192.168.255.255
-
NSM.External.Master.NIC
-
The physical or logical network interface for the external floating
address of the master node. There is no default value.
NSM.External.Master.NIC=hme0:3
-
NSM.External.Master.RouteAdd
-
Set the arguments to be passed to the route add
command when a master-eligible node becomes a master node. This parameter
is optional. If not defined, no routes are defined. There is no default value.
NSM.External.Master.RouteAdd=default 192.168.12.250
The preceding example generates the following command:
route add default 192.168.12.250
-
NSM.External.Master.RouteDelete
-
Set the arguments to be passed to the route delete command when a node loses the master role. This parameter is optional.
There is no default value.
NSM.External.Master.RouteDelete=default 192.168.12.250
The preceding example generates the following command:
route delete default 192.168.12.250
The following parameters are optional and must be configured only if
you have installed the Watchdog Timer on the nodes of your cluster.
-
WATCHDOG.NhasWatchdog
-
If this parameter is set to true, the product enables
the watchdog and initiates the patting of the watchdog.
Options are true or false. The
default is false.
WATCHDOG.NhasWatchdog=false
-
WATCHDOG.ShutDownTimeout
-
The maximum number of milliseconds to wait for the system to reach init run level 6 during a shutdown. See the init(1m) man page.
The value must be provided in milliseconds. The default value is 5000.
WATCHDOG.ShutDownTimeout=6000
-
WATCHDOG.OsTimeout
-
The maximum number of milliseconds to wait between the system boot at init run level 2 and init run level 6. See the init(1m)
man page.
The value must be provided in milliseconds. The default value is 3000.
WATCHDOG.OsTimeout=5000
-
WATCHDOG.PattingIntvl
-
The interval after which the hardware watchdog is re-enabled by the
Watchdog Timer daemon, nhwdtd. The value must be provided
in milliseconds and must be at least 1000 milliseconds less than the values
of the WATCHDOG.OsTimeout and WATCHDOG.ShutDownTimeout parameters. The default value is 2000.
patting_intvl=5000
Example 1. Example of the nhfs.conf File on a Master-Eligible
Node
### Common part
Node.DomainId=69
Node.NIC0=hme0:1
Node.NIC1=hme1
Node.NICCGTP=cgtp0
Node.UseCGTP=True
Node.type=Diskfull
### Cluster part
Cluster.Master.Id=1
### CMM part
CMM.IsEligible=True
CMM.LocalConfig.Dir=/etc/opt/SUNWcgha
### RNFS part
RNFS.StatdAlternatePath=/SUNWcgha/local
RNFS.Slice.0=/dev/rdsk/c0t0d0s3 /dev/rdsk/c0t0d0s5 /dev/rdsk/c0t0d0s3 \
/dev/rdsk/c0t0d0s5 1
RNFS.Slice.1=/dev/rdsk/c0t0d0s4 /dev/rdsk/c0t0d0s6 /dev/rdsk/c0t0d0s4 \
/dev/rdsk/c0t0d0s6 1
RNFS.Share.0=share -F nfs -o rw,root=b14-ct400-10-cgtp:b14-ct400-20-cgtp \
-d "SUNWcgha" /SUNWcgha/local/export
RNFS.Share.1=share -F nfs -o rw -d "Export" /export
RNFS.Installed=True
### NSM Part
NSM.Exec.MasterDir=/opt/SUNWcgha/actions/master
NSM.Exec.ViceMasterDir=/opt/SUNWcgha/actions/vicemaster
NSM.Log.MasterDir=/var/run/SUNWcgha/actions/master
NSM.Log.ViceMasterDir=/var/run/SUNWcgha/actions/vicemaster
NSM.External.Master.Address=192.168.12.39
NSM.External.Master.NIC=hme0:3
NSM.External.Master.Broadcast=192.168.255.255
NSM.External.Master.RouteAdd=default 192.168.12.250
NSM.External.Master.RouteDelete=default 192.168.12.250
Example 2. Example of the nhfs.conf File on a Diskless Node
### Common part
Node.DomainId=250
Node.NIC0=hme0:1
Node.NIC1=hme1
Node.NICCGTP=cgtp0
Node.UseCGTP=True
Node.type=Diskless
### CMM part
CMM.IsEligible=False
CMM.LocalConfig.Dir=/etc/opt/SUNWcgha
CMM.CurrentNodeName=netraDISKLESS1
Example 3. Example of the nhfs.conf File on a Dataless Node
Node.NICCGTP=cgtp0
Node.UseCGTP=True
Node.NIC1=eri1
Node.NIC0=eri0
Node.DomainId=250
Node.Type=Dataless
CMM.IsEligible=False
CMM.LocalConfig.Dir=/etc/opt/SUNWcgha
CMM.CurrentNodeName=netraDATALESS1
See attributes(5)
for descriptions of the following attributes:
ATTRIBUTE TYPE | ATTRIBUTE VALUE |
Architecture | SPARC |
Availability | SUNWnhcdt |
Interface Stability | Evolving |
init(1m), nhcrfsd(1M), nhcmmd(1M), share(1m), nhenablesync(1m), nhpmd(1M), nhnsmd(1M), cluster_nodes_table(4)
Netra HAS FS 2.1 | Go To Top | Last Changed September 2004 |
Company Info
|
Contact
|
Copyright 2004 Sun Microsystems, Inc. All rights reserved.
Use is subject to license terms.
|