How to Remove a Node From a Disk Device Group (Solstice DiskSuite/Solaris Volume Manager)
Use this procedure to remove a cluster node from the list of potential primaries of a Solstice DiskSuite/Solaris Volume Manager disk device group. Repeat the metaset command for each disk device group from which you want to remove the node.
Verify that the node is still a member of the group and that the group is an SDS/SVM device group.
Device group type SDS/SVM indicates a Solstice DiskSuite/Solaris Volume Manager disk device group.
phys-suncluster-1% scconf -pv | grep '(global-galileo)' (global-galileo) Device group type: SDS/SVM (global-galileo) Device group failback enabled: no (global-galileo) Device group node list: phys-suncluster-1, phys-suncluster-2 (global-galileo) Diskset name: global-galileo phys-suncluster-1%
Determine which node is the current primary for the device group.
# scstat -D
Become superuser on the node that currently owns the disk device group that you want to modify.
Delete the node's hostname from the disk device group.
# metaset -s setname -d -h nodelist
-s setname Specifies the disk device group name
-d Deletes from the disk device group the nodes identified with -h
-h nodelist Removes the node from the list of nodes that can master the disk device group
Note - The update can take several minutes to complete.
If the command fails, add the -f (Force) option to the command.
# metaset -s setname -d -f -h nodelist
Repeat Step 4 for each disk device group from which the node is being removed as a potential primary.
Verify that the node has been removed from the disk device group.
The disk device group name matches the diskset name that is specified with metaset.
phys-suncluster-1% scconf -pv |grep Device group node list: phys-suncluster-1, phys-suncluster-2, phys-suncluster-1%
Example--Removing a Node From a Disk Device Group (Solstice DiskSuite/Solaris Volume Manager)
The following example shows the removal of the host name phys-schost-2 from a disk device group configuration. This example eliminates phys-schost-2 as a potential primary for the designated disk device group. Verify removal of the node by running the scstat -D command. Check that the removed node is no longer displayed in the screen text.
[Determine the Solstice DiskSuite/Solaris Volume Manager disk device group(2) for the node:] # scconf -pv | grep Device Device group name: dg-schost-1 Device group type: SDS/SVM Device group failback enabled: no Device group node list: phys-schost-1, phys-schost-2 Device group ordered node list: yes Device group diskset name: dg-schost-1 [Determine the disk device group(s) for the node:] # scstat -D -- Device Group Servers -- Device Group Primary Secondary ------------ ------- --------- Device group servers: dg-schost-1 phys-schost-1 phys-schost-2 [Become superuser.] [Remove the hostname from the disk device group:] # metaset -s dg-schost-1 -d -h phys-schost-2 [Verify removal of the node:] phys-suncluster-1% scconf -pv |grep Device Group Servers -- Device Group Primary Secondary ------------ ------- --------- Device group node list: dg-schost-1, phys-schost-2, |
How to Create More Than Three Disksets in a Cluster
If you intend to create more than three disksets in the cluster, perform the following steps before you create the disksets. Follow these steps if you are installing disksets for the first time or if you are adding more disksets to a fully configured cluster.
Ensure that the value of the md_nsets variable is high enough. The value should accommodate the total number of disksets you intend to create in the cluster.
On any node of the cluster, check the value of the md_nsets variable in the /kernel/drv/md.conf file.
If the number of disksets in the cluster will be greater than the existing value of md_nsets minus one, increase the value of md_nsets on each node.
The maximum permissible number of disksets is the value of md_nsets minus one. The maximum possible value of md_nsets is 32.
Ensure that the /kernel/drv/md.conf file is identical on each node of the cluster.
Caution - Failure to follow this guideline can result in serious Solstice DiskSuite/Solaris Volume Manager
errors and possible loss of data.From one node, shut down the cluster.
# scshutdown -g0 -y
Reboot each node of the cluster.
ok> boot
On each node in the cluster, run the devfsadm(1M) command.
You can run this command on all nodes in the cluster at the same time.
From one node of the cluster, run the scgdevs(1M) command.
On each node, verify that the scgdevs command has completed before you attempt to create any disksets.
The scgdevs(1M) command calls itself remotely on all nodes, even when the command is run from just one node. To determine whether the scgdevs command has completed processing, run the following command on each node of the cluster.
% ps -ef | grep scgdevs
How to Create a New Disk Group When Initializing Disks (VERITAS Volume Manager)
Note - This procedure is only for initializing disks. If you are encapsulating disks, use the procedure How to Create a New Disk Group When Encapsulating Disks (VERITAS Volume Manager).
After adding the VxVM disk group, you need to register the disk device group.
If you use VxVM to set up shared disk groups for Oracle Parallel Server/Real Application Clusters, use the cluster functionality of VxVM as described in the VERITAS Volume Manager Administrator's Reference Guide.
Become superuser on any cluster node that is physically connected to disks that make up the disk group being added.
Create the VxVM disk group and volume.
Use your preferred method to create the disk group and volume.
Note - If you are setting up a mirrored volume, use Dirty Region Logging (DRL) to decrease volume recovery time after a node failure. However, DRL might decrease I/O throughput.
See the VERITAS Volume Manager documentation for the procedures to complete this step.
Register the VxVM disk group as a Sun Cluster disk device group.
See How to Register a Disk Group as a Disk Device Group (VERITAS Volume Manager).
Do not register the Oracle Parallel Server/Real Application Clusters shared disk groups with the cluster framework.