How to configure I/O Fencing

Problem

How to configure I/O Fencing

Solution

     1.       Verify the SCSI3 PR support for disks (coordinator and data), using

       vxfentsthdw utility with –r option, which is used as a read only.

a.       vxfentsthdw –r –g temp_DG

     2.       Initialize the disks to be used for coordinator DG.

a.       vxdisksetup –i disk1

b.      vxdisksetup –i disk2

c.       vxdisksetup –i disk3

     3.       Configure coordinated DG

a.       vxdg init vxfencoorddg disk1 disk2 disk3

b.      vxdg –g vxfencoorddg set coordinator=on

     4.       Deport DG, Import it again and deport it again.

a.       vxdg deport vxfencoorddg

b.      vxdg –t import vxfencoorddg  : -t flag is used to disable the autoimport of the DG at the next reboot.

c.       vxdg deport vxfencoorddg

     5.       Create /etc/vxfendg on all cluster nodes.

a.       This file must contain the coordinator disk group name.

b.      cp “vxfencoorddg” > /etc/vxfendg

     6.       Create /etc/vxfenmode on all cluster nodes.

a.       It will contain mode of fencing

b.      cp /etc/vxfen.d/vxfenmode_scsi3_dmp               /etc/vxfenmode

     7.       Start the fencing driver on each cluster node.

a.       /sbin/vxfen-startup

     8.       Copy main.cf file and then Save and close the configuration

a.       haconf –dump –makero

     9.       Set UseFence=SCSI3 in main.cf manually. Verify the main.cf using haconf –verify.

     10.   Stop VCS on all systems bringing all SGs offline.

a.       hatop –all

     11.   Restart VCS on all nodes

a.       hastart

Terms of use for this information are found in Legal Notices.