To manually set up I/O fencing in a non-SCSI-3 PR compliant setup

1. Configure I/O fencing in customized mode with only CP servers as coordination points.

   See Setting up server-based I/O fencing manually.

2. Make sure that the VCS cluster is online and check that the fencing mode is customized.

   # vxfenadm -d

3. Make sure that the cluster attribute UseFence is set to SCSI3.

   # haclus -value UseFence

4. On each node, edit the /etc/vxenviron file as follows:

   	data_disk_fencing=off

5. Enter the following command to change the vxfen_min_delay parameter value:

   # /usr/sbin/kctune vxfen_vxfnd_tmt=25

6. On each node, edit the /etc/vxfenmode file as follows:

   	loser_exit_delay=55
   	vxfen_script_timeout=25

   Refer to the sample /etc/vxfenmode file.

7. On each node, set the value of the LLT sendhbcap timer parameter value as follows:

   • Run the following command:

     lltconfig -T sendhbcap:3000

   • Add the following line to the /etc/llttab file so that the changes remain persistent after any reboot:

     set-timer senhbcap:3000

8. On any one node, edit the VCS configuration file as follows:

   • Make the VCS configuration file writable:

     	# haconf -makerw

   • For each resource of the type DiskGroup, set the value of the MonitorReservation attribute to 0 and the value of the Reservation attribute to NONE.

     	# hares -modify <dg_resource> MonitorReservation 0

     	# hares -modify <dg_resource> Reservation "NONE"

   • Run the following command to verify the value:

     	# hares -list Type=DiskGroup MonitorReservation!=0

      # hares -list Type=DiskGroup Reservation!="NONE"

     The command should not list any resources.

   • Modify the default value of the Reservation attribute at type-level.

      # haattr -default DiskGroup Reservation "NONE"

   • Make the VCS configuration file read-only

     	# haconf -dump -makero

9. Make sure that the UseFence attribute in the VCS configuration file main.cf is set to SCSI3.

10. To make these VxFEN changes take effect, stop and restart VxFEN and the dependent modules

    • On each node, run the following command to stop VCS:

      /sbin/init.d/vcs stop

    • After VCS takes all services offline, run the following command to stop VxFEN:

      /sbin/init.d/vxfen stop

    • On each node, run the following commands to restart VxFEN and VCS:

       # /sbin/init.d/vxfen start
      	# /sbin/init.d/vcs start