The following are highlights of the procedure to modify the existing VCS configuration on the secondary site:
Add the log owner and RVG service groups.
Add a service group to manage the application database and the supporting resources.
Define the replication objects and agents, such that the cluster at the secondary site can function as a companion to the primary cluster.
The following steps are similar to those performed on the primary site.
Note: |
The example precedure illustrates the confguration process using a manual file editing method. If you are using the Java Console, some steps do not apply in the same order. |
To modify VCS on the secondary site
# haconf -dump -makero
# cd /etc/VRTSvcs/conf/config # cp main.cf main.orig
Review the sample configuration file after the SFCFSHA installation to see the CVM configuration.
In our example, the secondary site has clus2 consisting of the nodes mercury and jupiter. To modify the CVM service group on the secondary site, use the CVM group on the primary site as your guide.
Add a failover service group using the appropriate values for your cluster and nodes. Include the following resources:
RVGLogowner resource. The node on which the group is online functions as the log owner (node connected to the second cluster for the purpose of replicating data).
IP resource
NIC resources
group rlogowner ( SystemList = { mercury = 0, jupiter = 1 } AutoStartList = { mercury, jupiter } ) IP logowner_ip ( Device = bge0 Address = "10.11.9.102" NetMask = "255.255.255.0" )
NIC nic ( Device = bge0 NetworkHosts = { "10.10.8.1" } NetworkType = ether )
RVGLogowner logowner ( RVG = rac1_rvg DiskGroup = oradatadg )
requires group RVGgroup online local firm logowner requires logowner_ip logowner_ip requires nic
The following is an example RVGgroup service group:
group RVGgroup ( SystemList = { mercury = 0, jupiter = 1 } Parallel = 1 AutoStartList = { mercury, jupiter } ) RVGShared racdata_rvg ( RVG = rac1_rvg DiskGroup = oradatadg ) CVMVolDg racdata_voldg ( CVMDiskGroup = oradatadg CVMActivation = sw ) requires group cvm online local firm racdata_rvg requires racdata_voldg
Add an application service group. Use the application service group on the primary site as a model for the application service group on the secondary site.
Define the application service group as a global group by specifying the clusters on the primary and secondary sites as values for the ClusterList group attribute.
Assign this global group the same name as the group on the primary site; for example, database_grp.
Include the ClusterList and ClusterFailOverPolicy cluster attributes. Symantec recommends using the Manual value.
Add the RVGSharedPri resource to the group configuration.
Remove the CVMVolDg resource, if it has been configured in your previous configuration. This resource is now part of the RVG service group.
Specify the service group to depend (online, local, firm) on the RVG service group.
group database_grp ( SystemList = { mercury = 0, jupiter = 1 } ClusterList = { clus2 = 0, clus1 = 1 } Parallel = 1 OnlineRetryInterval = 300 ClusterFailOverPolicy = Manual Authority = 1 AutoStartList = { mercury, jupiter } ) RVGSharedPri ora_vvr_shpri ( RvgResourceName = racdata_rvg OnlineRetryLimit = 0 ) CFSMount oradata_mnt ( MountPoint = "/oradata" BlockDevice = "/dev/vx/dsk/oradatadg/racdb_vol" Critical = 0 )
Process vxfend ( PathName = "/sbin/vxfend" Arguments = "-m sybase -k /tmp/vcmp_socket" ) RVGSharedPri ora_vvr_shpri ( RvgResourceName = racdata_rvg OnlineRetryLimit = 0 ) requires group RVGgroup online local firm oradata_mnt requires ora_vvr_shpri
# hacf -verify /etc/VRTSvcs/conf/config
# hastop -all -force
Wait for port h to stop on all nodes, and then restart VCS with the new configuration on all primary nodes:
# hastart
# hagrp -display
The application, RVG, and CVM groups are online on both nodes of the primary site. The RVGLogOwner and ClusterService groups are online on one node of the cluster. If either the RVG group or the RVGLogOwner group is partially online, manually bring the groups online using the hagrp -online command. This information applies to the secondary site, except for the application group which must be offline.
# hagrp -display
The application service group is offline on the secondary site, but the ClusterService, CVM, RVG log owner, and RVG groups are online.
This completes the setup for an SFCFSHA global cluster using VVR for replication. Symantec recommends testing a global cluster before putting it into production.