About setting up I/O fencing


About setting up I/O fencing
Prev	Configuring VCS clusters for data integrity	Next

Figure: Workflow to configure I/O fencing illustrates the tasks involved to configure I/O fencing.

Figure: Workflow to configure I/O fencing

Workflow to configure disk-based I/O fencing

I/O fencing requires the coordinator disks be configured in a disk group. The coordinator disks must be accessible to each node in the cluster. These disks enable the vxfen driver to resolve potential split-brain conditions and prevent data corruption.

Review the following requirements for coordinator disks:

You must have three coordinator disks.
Each of the coordinator disks must use a physically separate disk or LUN.
Each of the coordinator disks should exist on a different disk array, if possible.
You must initialize each disk as a VxVM disk.
The coordinator disks must support SCSI-3 persistent reservations.
The coordinator disks must exist in a disk group (for example, vxfencoorddg).
Symantec recommends using hardware-based mirroring for coordinator disks.

The I/O fencing configuration files include:

/etc/vxfendg

You must create this file to include the coordinator disk group information.

/etc/vxfenmode

You must set the I/O fencing mode to SCSI-3.

You can configure the vxfen module to use either DMP devices or the underlying raw character devices. Note that you must use the same SCSI-3 disk policy on all the nodes. The SCSI-3 disk policy can either be raw or dmp. The policy is raw by default.

/etc/vxfentab

When you run the vxfen startup file to start I/O fencing, the script creates this /etc/vxfentab file on each node with a list of all paths to each coordinator disk. The startup script uses the contents of the /etc/vxfendg and /etc/vxfenmode files.

Thus any time a system is rebooted, the fencing driver reinitializes the vxfentab file with the current list of all paths to the coordinator disks.

Note:

The /etc/vxfentab file is a generated file; do not modify this file.

An example of the /etc/vxfentab file on one node resembles as follows:

Raw disk:

/dev/rdsk/c1t1d0s2
/dev/rdsk/c2t1d0s2
/dev/rdsk/c3t1d2s2

DMP disk:

/dev/vx/rdmp/c1t1d0s2
/dev/vx/rdmp/c2t1d0s2
/dev/vx/rdmp/c3t1d0s2

In some cases you must remove disks from or add disks to an existing coordinator disk group.

Warning:

If you remove disks from an existing coordinator disk group, then be sure to remove the registration and reservation keys from these disks before you add the disks to another disk group.

Prev	Up	Next
About coordination points	Home	Preparing to configure I/O fencing