Uinstall the Veritas Cluster Service .

Below are the general outline for uninstall veritas cluster services without removing Veritas Volume manger and file system . The below stpes need to perform in both the cluster nodes . Please perfrom the steps in one node at a time.

• Make a backup of the VCS configuration file, to reference for file systems, shares and apps if needed

o cp /etc/VRTSvcs/conf/config/main.cf $HOME/main.cf

• Stop VCS, but keep all applications running

o hastop -all -force

• Disable VCS startup

o mv /etc/rc3.d/S99vcs /etc/rc3.d/s99vcs
o mv /etc/rc2.d/S92gab /etc/rc2.d/s92gab
o mv /etc/rc2.d/S92gab /etc/rc2.d/s92gab
o rm /etc/llttab
o rm /etc/gabtab
o rm /etc/llthost

• Add all file systems to /etc/vfstab

• Add Solaris Services back that were removed per VCS requirement

o /network/nfs/status
o /network/nfs/server
o /network/nfs/mapid
o Here are the commands used to remove those services:
 svccfg delete -f svc:/network/nfs/status:default
 svccfg delete -f svc:/network/nfs/server:default
 svccfg delete -f svc:/network/nfs/mapid:default
o The service manifest files to import should be located in /var/svc/manifest/network/nfs
o Below steps will import the manifest (You will get the error that “ partial import because of default-milestone is already online” while importing the nfs server manifest , but it will come online after server reboot)
 svccfg
 svc:> validate /var/svc/manifest/network/nfs/server.xml
 svc:> import /var/svc/manifest/network/nfs/server.xml

 svc:> validate /var/svc/manifest/network/nfs/status.xml
 svc:> import /var/svc/manifest/network/nfs/status.xml

 svc:> validate /var/svc/manifest/network/nfs/mapid.xml
 svc:> import /var/svc/manifest/network/nfs/mapid.xml

• Configure NFS shares in /etc/dfs/dfstab

• Enable the nfs/server service (and all dependency services)

• Ensure NFS shares are shared

o May need to log into every server that mounts these shares to verify they are still mounted, and remount if they are not.

• Configure IPMP
o Generally, the Primary Interface is “ce0” and the secondary interface is “ce4”, but verify within the main.cf file, looking for the MultiNICA definition stanza.

• Ensure all apps that were managed by VCS (Oracle, SAP, etc) have the appropriate startup/shutdown scripts in system startup/shutdown. The app teams will need to write the startup/shutdown scripts.
o Ensure you link the shutdown scripts into every rc directory except rc3.d (app teams typically state to only put into one rc directory, but that is incorrect for Solaris)

• Reboot the servers and verify everything comes up appropriately

o IPMP
o File systems
o NFS Shares
o Apps

• May need to log into every server that mounts these shares to verify they are still mounted, and remount if they are not.

• Once verified, uninstall VCS
o Not VxVM or VxFS as those are still used, just the VCS components. This should be the complete list, but verify:
 VRTSsap
 VRTSvcs
 VRTSvcsag
 VRTSvcsdc
 VRTSvcsmg
 VRTSvcsmn
 VRTSvcsor
 VRTSvcsvr
 VRTSvcsw
 VRTScscw
 VRTScsocw
 VRTSagtfw

• Reboot the servers again with VCS uninstalled, ensuring proper startup

• May need to log into every server that mounts these shares to verify they are still mounted, and remount if they are not.



Some of the work can be done before the actual scheduled maintenance window:

• Backup copy of the VCS configuration file

• Add all file systems to /etc/vfstab, though commented out in the event the server is rebooted before the maintenance window

• Configure NFS shares in /etc/dfs/dfstab (though cannot add the services back yet)

• Obtain the needed IP Addresses to properly configure IPMP

• Pre-create the new /etc/hostname.* files, though naming them differently in case the server is rebooted before the maintenance window

• Place the app startup/shutdown scripts into place, though naming their links to ensure they do not run on server startup/shutdown in the event the server is rebooted before the maintenance window